Compare commits
6
Commits
| Author | SHA1 | Date | |
|---|---|---|---|
 sysops
|
c4840ca55b | ||
|
sysops
|
34a705390c | ||
|
sysops
|
627961b972 | ||
|
sysopsandClaude Sonnet 5
|
27f9866264 | ||
|
sysopsandClaude Sonnet 5
|
3d20d86a4f | ||
|
sysopsandClaude Sonnet 5
|
e4793303fc |
@@ -129,34 +129,3 @@ Keine Commits in dieser Session.
|
||||
- migrations/0001_tenant_registry.sql | 10 ++++++++++
|
||||
|
||||
---
|
||||
## 2026-08-28 23:55 – 23:59 (3m)
|
||||
**Beschreibung:** Claude Code Session
|
||||
**Projekt:** nexarch
|
||||
|
||||
### Commits
|
||||
- 369a40a OPS-05: alerting-bei-schwellwert-ueberschreitung (internal/alerting: regel-store, evaluator gegen ops-03-metriken, cfg-02-zustellung, drosselung je regel+zeitreihe)
|
||||
- 06dbd52 Merge branch 'feature/cfg-02-benachrichtigungs-dispatcher-core-service-fuer-module' into feature/ops-05-alerting-bei-schwellwert-ueberschreitung
|
||||
|
||||
### Geänderte Dateien
|
||||
- internal/alerting/evaluator.go | 194 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|
||||
- internal/alerting/evaluator_test.go | 233 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|
||||
- internal/alerting/rules.go | 132 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|
||||
- migrations/0006_alert_rules.down.sql | 2 ++
|
||||
- migrations/0006_alert_rules.up.sql | 24 ++++++++++++++++++++++
|
||||
|
||||
---
|
||||
## 2026-08-29 00:00 – 00:00 (0m)
|
||||
**Beschreibung:** Claude Code Session
|
||||
**Projekt:** code
|
||||
|
||||
### Commits
|
||||
Keine Commits in dieser Session.
|
||||
|
||||
### Geänderte Dateien
|
||||
- internal/alerting/evaluator.go | 194 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|
||||
- internal/alerting/evaluator_test.go | 233 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|
||||
- internal/alerting/rules.go | 132 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
|
||||
- migrations/0006_alert_rules.down.sql | 2 ++
|
||||
- migrations/0006_alert_rules.up.sql | 24 ++++++++++++++++++++++
|
||||
|
||||
---
|
||||
|
||||
+10
-1
@@ -8,6 +8,7 @@ import (
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/config"
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/db"
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/tenant"
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/user"
|
||||
)
|
||||
|
||||
func main() {
|
||||
@@ -34,12 +35,20 @@ func main() {
|
||||
provisioner := tenant.NewProvisioner(adminPool, registry, cfg.TenantDSNTemplate)
|
||||
tenantHandler := tenant.NewHandler(provisioner)
|
||||
|
||||
// Superadmin-Konten leben mandantenuebergreifend in der Registry-DB.
|
||||
// Tenant-User-CRUD (user.TenantUserStore) braucht Connection-Routing pro
|
||||
// Mandant (TEN-06, noch nicht gebaut) und wird hier bewusst noch nicht
|
||||
// verdrahtet — Package ist bereits eigenstaendig nutzbar/testbar.
|
||||
superadmins := user.NewSuperadminStore(registryPool)
|
||||
userHandler := user.NewHandler(nil, superadmins)
|
||||
|
||||
mux := http.NewServeMux()
|
||||
mux.HandleFunc("/healthz", func(w http.ResponseWriter, r *http.Request) {
|
||||
w.WriteHeader(http.StatusOK)
|
||||
})
|
||||
// Vorlaeufiger Pfad ohne Versionierung/Auth — wird mit API-01/IAM-01 abgeloest.
|
||||
// Vorlaeufige Pfade ohne Versionierung/Auth — werden mit API-01/IAM-02 abgeloest.
|
||||
mux.HandleFunc("/internal/tenants", tenantHandler.CreateTenant)
|
||||
mux.HandleFunc("/internal/superadmins", userHandler.CreateSuperadmin)
|
||||
|
||||
log.Printf("nexarch-core listening on %s", cfg.ListenAddr)
|
||||
if err := http.ListenAndServe(cfg.ListenAddr, mux); err != nil {
|
||||
|
||||
@@ -1,43 +0,0 @@
|
||||
// metrics-devserver stellt den OPS-03-Metrics-Aggregator (internal/metrics)
|
||||
// unter /metrics bereit, damit ein echter Prometheus-Scrape-Vorgang gegen
|
||||
// den Core-Dienst geprueft werden kann (Pruefung 3). Getrennt von cmd/core
|
||||
// aus demselben Grund wie die anderen *-devserver.
|
||||
package main
|
||||
|
||||
import (
|
||||
"context"
|
||||
"log"
|
||||
"net/http"
|
||||
"os"
|
||||
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/db"
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/metrics"
|
||||
)
|
||||
|
||||
func main() {
|
||||
dsn := os.Getenv("NEXARCH_REGISTRY_DSN")
|
||||
if dsn == "" {
|
||||
log.Fatal("NEXARCH_REGISTRY_DSN nicht gesetzt")
|
||||
}
|
||||
addr := os.Getenv("NEXARCH_METRICS_LISTEN_ADDR")
|
||||
if addr == "" {
|
||||
addr = ":8085"
|
||||
}
|
||||
|
||||
ctx := context.Background()
|
||||
pool, err := db.Connect(ctx, dsn)
|
||||
if err != nil {
|
||||
log.Fatalf("db: %v", err)
|
||||
}
|
||||
defer pool.Close()
|
||||
|
||||
sourceStore := metrics.NewSourceStore(pool)
|
||||
agg := metrics.NewAggregator(metrics.NewCoreRegistry(), sourceStore.Provide)
|
||||
|
||||
mux := http.NewServeMux()
|
||||
mux.HandleFunc("/metrics", agg.Handler())
|
||||
mux.HandleFunc("/healthz", func(w http.ResponseWriter, r *http.Request) { w.WriteHeader(http.StatusOK) })
|
||||
|
||||
log.Printf("metrics-devserver listening on %s", addr)
|
||||
log.Fatal(http.ListenAndServe(addr, mux))
|
||||
}
|
||||
@@ -1,25 +1,17 @@
|
||||
module gitea.perlbach24.de/scripte/nexarch
|
||||
|
||||
go 1.25.0
|
||||
go 1.22
|
||||
|
||||
require (
|
||||
github.com/golang-jwt/jwt/v5 v5.3.1
|
||||
github.com/jackc/pgx/v5 v5.6.0
|
||||
github.com/prometheus/client_golang v1.24.1
|
||||
github.com/prometheus/client_model v0.6.2
|
||||
github.com/prometheus/common v0.70.1
|
||||
golang.org/x/crypto v0.17.0
|
||||
)
|
||||
|
||||
require (
|
||||
github.com/beorn7/perks v1.0.1 // indirect
|
||||
github.com/cespare/xxhash/v2 v2.3.0 // indirect
|
||||
github.com/jackc/pgpassfile v1.0.0 // indirect
|
||||
github.com/jackc/pgservicefile v0.0.0-20221227161230-091c0ba34f0a // indirect
|
||||
github.com/jackc/puddle/v2 v2.2.1 // indirect
|
||||
github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 // indirect
|
||||
github.com/prometheus/procfs v0.21.1 // indirect
|
||||
golang.org/x/crypto v0.17.0 // indirect
|
||||
golang.org/x/sync v0.22.0 // indirect
|
||||
golang.org/x/sys v0.47.0 // indirect
|
||||
golang.org/x/text v0.40.0 // indirect
|
||||
google.golang.org/protobuf v1.36.11 // indirect
|
||||
golang.org/x/sync v0.1.0 // indirect
|
||||
golang.org/x/text v0.14.0 // indirect
|
||||
)
|
||||
|
||||
@@ -1,12 +1,8 @@
|
||||
github.com/beorn7/perks v1.0.1 h1:VlbKKnNfV8bJzeqoa4cOKqO6bYr3WgKZxO8Z16+hsOM=
|
||||
github.com/beorn7/perks v1.0.1/go.mod h1:G2ZrVWU2WbWT9wwq4/hrbKbnv/1ERSJQ0ibhJ6rlkpw=
|
||||
github.com/cespare/xxhash/v2 v2.3.0 h1:UL815xU9SqsFlibzuggzjXhog7bL6oX9BbNZnL2UFvs=
|
||||
github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
|
||||
github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
|
||||
github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
|
||||
github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
|
||||
github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
|
||||
github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
|
||||
github.com/golang-jwt/jwt/v5 v5.3.1 h1:kYf81DTWFe7t+1VvL7eS+jKFVWaUnK9cB1qbwn63YCY=
|
||||
github.com/golang-jwt/jwt/v5 v5.3.1/go.mod h1:fxCRLWMO43lRc8nhHWY6LGqRcf+1gQWArsqaEUEa5bE=
|
||||
github.com/jackc/pgpassfile v1.0.0 h1:/6Hmqy13Ss2zCq62VdNG8tM1wchn8zjSGOBJ6icpsIM=
|
||||
github.com/jackc/pgpassfile v1.0.0/go.mod h1:CEx0iS5ambNFdcRtxPj5JhEz+xB6uRky5eyVu/W2HEg=
|
||||
github.com/jackc/pgservicefile v0.0.0-20221227161230-091c0ba34f0a h1:bbPeKD0xmW/Y25WS6cokEszi5g+S0QxI/d45PkRi7Nk=
|
||||
@@ -15,37 +11,19 @@ github.com/jackc/pgx/v5 v5.6.0 h1:SWJzexBzPL5jb0GEsrPMLIsi/3jOo7RHlzTjcAeDrPY=
|
||||
github.com/jackc/pgx/v5 v5.6.0/go.mod h1:DNZ/vlrUnhWCoFGxHAG8U2ljioxukquj7utPDgtQdTw=
|
||||
github.com/jackc/puddle/v2 v2.2.1 h1:RhxXJtFG022u4ibrCSMSiu5aOq1i77R3OHKNJj77OAk=
|
||||
github.com/jackc/puddle/v2 v2.2.1/go.mod h1:vriiEXHvEE654aYKXXjOvZM39qJ0q+azkZFrfEOc3H4=
|
||||
github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 h1:C3w9PqII01/Oq1c1nUAm88MOHcQC9l5mIlSMApZMrHA=
|
||||
github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822/go.mod h1:+n7T8mK8HuQTcFwEeznm/DIxMOiR9yIdICNftLE1DvQ=
|
||||
github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
|
||||
github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
|
||||
github.com/prometheus/client_golang v1.24.1 h1:JnJkREXzWxUdCuPFpIWZiPispT9xVV59uiuyR2bPlnU=
|
||||
github.com/prometheus/client_golang v1.24.1/go.mod h1:F+oSRECHg4sse5ucfYpYDeIv/hu68Zo0uoHKetWnzcE=
|
||||
github.com/prometheus/client_model v0.6.2 h1:oBsgwpGs7iVziMvrGhE53c/GrLUsZdHnqNwqPLxwZyk=
|
||||
github.com/prometheus/client_model v0.6.2/go.mod h1:y3m2F6Gdpfy6Ut/GBsUqTWZqCUvMVzSfMLjcu6wAwpE=
|
||||
github.com/prometheus/common v0.70.1 h1:1HvjP4D5oL3t8RsPlwxA9onvvStjtIHYE5XuuwOi/PY=
|
||||
github.com/prometheus/common v0.70.1/go.mod h1:VdFUQDMZK3VLkurFUVhia6uys/0suUp86TJz5qbJRhc=
|
||||
github.com/prometheus/procfs v0.21.1 h1:GljZCt+zSTS+NZq88cyQ1LjZ+RCHp3uVuabBWA5+OJI=
|
||||
github.com/prometheus/procfs v0.21.1/go.mod h1:aB55Cww9pdSJVHk0hUf0inxWyyjPogFIjmHKYgMKmtY=
|
||||
github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
|
||||
github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UVUgZn+9EI=
|
||||
github.com/stretchr/testify v1.7.0/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
|
||||
github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U=
|
||||
github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
|
||||
go.uber.org/goleak v1.3.0 h1:2K3zAYmnTNqV73imy9J1T3WC+gmCePx2hEGkimedGto=
|
||||
go.uber.org/goleak v1.3.0/go.mod h1:CoHD4mav9JJNrW/WLlf7HGZPjdw8EucARQHekz1X6bE=
|
||||
go.yaml.in/yaml/v2 v2.4.4 h1:tuyd0P+2Ont/d6e2rl3be67goVK4R6deVxCUX5vyPaQ=
|
||||
go.yaml.in/yaml/v2 v2.4.4/go.mod h1:gMZqIpDtDqOfM0uNfy0SkpRhvUryYH0Z6wdMYcacYXQ=
|
||||
github.com/stretchr/testify v1.8.1 h1:w7B6lhMri9wdJUVmEZPGGhZzrYTPvgJArz7wNPgYKsk=
|
||||
github.com/stretchr/testify v1.8.1/go.mod h1:w2LPCIKwWwSfY2zedu0+kehJoqGctiVI29o6fzry7u4=
|
||||
golang.org/x/crypto v0.17.0 h1:r8bRNjWL3GshPW3gkd+RpvzWrZAwPS49OmTGZ/uhM4k=
|
||||
golang.org/x/crypto v0.17.0/go.mod h1:gCAAfMLgwOJRpTjQ2zCCt2OcSfYMTeZVSRtQlPC7Nq4=
|
||||
golang.org/x/sync v0.22.0 h1:SZjpbeLmrCk4xhRSZFNZW5gFUeCeFgjekvI/+gfScek=
|
||||
golang.org/x/sync v0.22.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
|
||||
golang.org/x/sys v0.47.0 h1:o7XGOvZQCADBQQ4Y7VNq2dRWQR7JmOUW8Kxx4ZsNgWs=
|
||||
golang.org/x/sys v0.47.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
|
||||
golang.org/x/text v0.40.0 h1:Ub2Z6/xjgF1WrYQz2nuITOEegKFtiIy+rieRJ5lHZKs=
|
||||
golang.org/x/text v0.40.0/go.mod h1:hpnzDAfGV753zIKo+wk3u1bVKCGPbrnF7+7LBF/UHVY=
|
||||
google.golang.org/protobuf v1.36.11 h1:fV6ZwhNocDyBLK0dj+fg8ektcVegBBuEolpbTQyBNVE=
|
||||
google.golang.org/protobuf v1.36.11/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=
|
||||
golang.org/x/sync v0.1.0 h1:wsuoTGHzEhffawBOhz5CYhcrV4IdKZbEyZjBMuTp12o=
|
||||
golang.org/x/sync v0.1.0/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
|
||||
golang.org/x/text v0.14.0 h1:ScX5w1eTa3QqT8oi6+ziP7dTV1S2+ALU0bI+0zXKWiQ=
|
||||
golang.org/x/text v0.14.0/go.mod h1:18ZOQIKpY8NJVqYksKHtTdi31H5itFRjB5/qKTNYzSU=
|
||||
gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
|
||||
gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
|
||||
gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
|
||||
|
||||
@@ -1,194 +0,0 @@
|
||||
package alerting
|
||||
|
||||
import (
|
||||
"context"
|
||||
"fmt"
|
||||
"sort"
|
||||
"strings"
|
||||
"time"
|
||||
|
||||
dto "github.com/prometheus/client_model/go"
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/notify"
|
||||
)
|
||||
|
||||
// DefaultDebounceInterval: wiederholte Alarmierung für denselben
|
||||
// anhaltenden Zustand ist gedrosselt (Akzeptanzkriterium 3) — 15 Minuten
|
||||
// ist ein üblicher Kompromiss zwischen "schnell genug informiert" und
|
||||
// "kein Alarm-Spam bei dauerhaft überschrittenem Wert".
|
||||
const DefaultDebounceInterval = 15 * time.Minute
|
||||
|
||||
// AlertChannel ist der CFG-02-Kanal, über den Schwellwert-Alarme zugestellt
|
||||
// werden — ein eigener Kanalname, damit Zustellregeln/-vorlagen (CFG-03)
|
||||
// unabhängig von anderen Benachrichtigungsarten konfiguriert werden können.
|
||||
const AlertChannel = "alert"
|
||||
|
||||
// Evaluator prüft konfigurierte Regeln gegen aktuell gesammelte Metriken
|
||||
// (aus internal/metrics.Aggregator.Gather) und löst bei Überschreitung eine
|
||||
// Benachrichtigung über CFG-02 aus (Akzeptanzkriterium 2), gedrosselt je
|
||||
// Regel+Zeitreihe (Akzeptanzkriterium 3).
|
||||
type Evaluator struct {
|
||||
rules *RuleStore
|
||||
debounce *debounceStore
|
||||
dispatcher *notify.Dispatcher
|
||||
interval time.Duration
|
||||
}
|
||||
|
||||
func NewEvaluator(rules *RuleStore, dispatcher *notify.Dispatcher, debouncePool *pgxpool.Pool, interval time.Duration) *Evaluator {
|
||||
if interval <= 0 {
|
||||
interval = DefaultDebounceInterval
|
||||
}
|
||||
return &Evaluator{
|
||||
rules: rules,
|
||||
debounce: &debounceStore{pool: debouncePool},
|
||||
dispatcher: dispatcher,
|
||||
interval: interval,
|
||||
}
|
||||
}
|
||||
|
||||
// FiredAlert beschreibt einen tatsächlich ausgelösten (nicht gedrosselten)
|
||||
// Alarm — fürs Testen/Logging, nicht Teil des öffentlichen Zustellwegs.
|
||||
type FiredAlert struct {
|
||||
RuleID string
|
||||
MetricName string
|
||||
Value float64
|
||||
Threshold float64
|
||||
Labels map[string]string
|
||||
Skipped bool // true, wenn wegen Drosselung NICHT tatsaechlich zugestellt
|
||||
}
|
||||
|
||||
// Evaluate prüft alle konfigurierten Regeln gegen families (Akzeptanzkriterium 1).
|
||||
// Für jede Zeitreihe, die eine Regel verletzt, wird — sofern nicht gedrosselt
|
||||
// — eine Benachrichtigung mit Metrik/Wert/Schwellwert/Labels (Tenant/Modul,
|
||||
// falls als Label vorhanden) über CFG-02 eingereiht (Akzeptanzkriterium 2).
|
||||
func (e *Evaluator) Evaluate(ctx context.Context, families []*dto.MetricFamily) ([]FiredAlert, error) {
|
||||
rules, err := e.rules.ListRules(ctx)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("regeln laden: %w", err)
|
||||
}
|
||||
if len(rules) == 0 {
|
||||
return nil, nil
|
||||
}
|
||||
|
||||
byName := make(map[string]*dto.MetricFamily, len(families))
|
||||
for _, f := range families {
|
||||
if f.Name != nil {
|
||||
byName[*f.Name] = f
|
||||
}
|
||||
}
|
||||
|
||||
var fired []FiredAlert
|
||||
for _, rule := range rules {
|
||||
family, ok := byName[rule.MetricName]
|
||||
if !ok {
|
||||
continue // Metrik (noch) nicht vorhanden -> keine Aussage moeglich, kein Fehler.
|
||||
}
|
||||
|
||||
for _, m := range family.Metric {
|
||||
labels := labelMap(m)
|
||||
if !matchesFilter(labels, rule.LabelFilters) {
|
||||
continue
|
||||
}
|
||||
value, ok := metricValue(m)
|
||||
if !ok {
|
||||
continue
|
||||
}
|
||||
if !violates(rule, value) {
|
||||
continue
|
||||
}
|
||||
|
||||
ruleKey := ruleKeyFor(rule.ID, labels)
|
||||
allowed, err := e.debounce.shouldFire(ctx, ruleKey, e.interval.Seconds())
|
||||
if err != nil {
|
||||
return fired, fmt.Errorf("drosselung pruefen: %w", err)
|
||||
}
|
||||
|
||||
alert := FiredAlert{
|
||||
RuleID: rule.ID, MetricName: rule.MetricName, Value: value,
|
||||
Threshold: rule.Threshold, Labels: labels, Skipped: !allowed,
|
||||
}
|
||||
fired = append(fired, alert)
|
||||
|
||||
if !allowed {
|
||||
continue
|
||||
}
|
||||
|
||||
payload := map[string]any{
|
||||
"metric": rule.MetricName,
|
||||
"value": value,
|
||||
"threshold": rule.Threshold,
|
||||
"comparison": string(rule.Comparison),
|
||||
"description": rule.Description,
|
||||
"labels": labels,
|
||||
}
|
||||
if _, err := e.dispatcher.Enqueue(ctx, AlertChannel, rule.Recipient, payload); err != nil {
|
||||
return fired, fmt.Errorf("alarm einreihen: %w", err)
|
||||
}
|
||||
}
|
||||
}
|
||||
return fired, nil
|
||||
}
|
||||
|
||||
func violates(rule Rule, value float64) bool {
|
||||
switch rule.Comparison {
|
||||
case ComparisonGreaterThan:
|
||||
return value > rule.Threshold
|
||||
case ComparisonLessThan:
|
||||
return value < rule.Threshold
|
||||
default:
|
||||
return false
|
||||
}
|
||||
}
|
||||
|
||||
func labelMap(m *dto.Metric) map[string]string {
|
||||
out := make(map[string]string, len(m.Label))
|
||||
for _, l := range m.Label {
|
||||
if l.Name != nil && l.Value != nil {
|
||||
out[*l.Name] = *l.Value
|
||||
}
|
||||
}
|
||||
return out
|
||||
}
|
||||
|
||||
func matchesFilter(labels, filter map[string]string) bool {
|
||||
for k, v := range filter {
|
||||
if labels[k] != v {
|
||||
return false
|
||||
}
|
||||
}
|
||||
return true
|
||||
}
|
||||
|
||||
func metricValue(m *dto.Metric) (float64, bool) {
|
||||
switch {
|
||||
case m.Gauge != nil && m.Gauge.Value != nil:
|
||||
return *m.Gauge.Value, true
|
||||
case m.Counter != nil && m.Counter.Value != nil:
|
||||
return *m.Counter.Value, true
|
||||
case m.Untyped != nil && m.Untyped.Value != nil:
|
||||
return *m.Untyped.Value, true
|
||||
default:
|
||||
return 0, false
|
||||
}
|
||||
}
|
||||
|
||||
// ruleKeyFor macht die Drosselung unabhaengig je Regel UND je konkreter
|
||||
// Zeitreihe (z. B. verschiedene Tenants/Module derselben Metrik loesen
|
||||
// unabhaengig voneinander aus, siehe Migrationskommentar).
|
||||
func ruleKeyFor(ruleID string, labels map[string]string) string {
|
||||
keys := make([]string, 0, len(labels))
|
||||
for k := range labels {
|
||||
keys = append(keys, k)
|
||||
}
|
||||
sort.Strings(keys)
|
||||
var b strings.Builder
|
||||
b.WriteString(ruleID)
|
||||
for _, k := range keys {
|
||||
b.WriteString("|")
|
||||
b.WriteString(k)
|
||||
b.WriteString("=")
|
||||
b.WriteString(labels[k])
|
||||
}
|
||||
return b.String()
|
||||
}
|
||||
@@ -1,233 +0,0 @@
|
||||
package alerting
|
||||
|
||||
import (
|
||||
"context"
|
||||
"os"
|
||||
"strings"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
dto "github.com/prometheus/client_model/go"
|
||||
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/notify"
|
||||
)
|
||||
|
||||
func setupTest(t *testing.T) (*RuleStore, *notify.Dispatcher, *pgxpool.Pool, func()) {
|
||||
t.Helper()
|
||||
adminDSN := os.Getenv("TEST_ADMIN_DSN")
|
||||
if adminDSN == "" {
|
||||
t.Skip("TEST_ADMIN_DSN nicht gesetzt, Integrationstest uebersprungen")
|
||||
}
|
||||
ctx := context.Background()
|
||||
|
||||
pool, err := pgxpool.New(ctx, adminDSN)
|
||||
if err != nil {
|
||||
t.Fatalf("pool: %v", err)
|
||||
}
|
||||
if _, err := pool.Exec(ctx, `
|
||||
CREATE TABLE IF NOT EXISTS alert_rules (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(), metric_name TEXT NOT NULL,
|
||||
comparison TEXT NOT NULL CHECK (comparison IN ('gt','lt')), threshold DOUBLE PRECISION NOT NULL,
|
||||
label_filters JSONB NOT NULL DEFAULT '{}'::jsonb, recipient TEXT NOT NULL,
|
||||
description TEXT NOT NULL DEFAULT '', created_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
CREATE TABLE IF NOT EXISTS alert_debounce_state (
|
||||
rule_key TEXT PRIMARY KEY, last_fired_at TIMESTAMPTZ NOT NULL
|
||||
);
|
||||
CREATE TABLE IF NOT EXISTS notification_jobs (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(), channel TEXT NOT NULL, recipient TEXT NOT NULL,
|
||||
payload JSONB NOT NULL DEFAULT '{}'::jsonb, status TEXT NOT NULL DEFAULT 'pending', attempts INT NOT NULL DEFAULT 0,
|
||||
max_attempts INT NOT NULL DEFAULT 5, next_attempt_at TIMESTAMPTZ NOT NULL DEFAULT now(), last_error TEXT,
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now(), updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
`); err != nil {
|
||||
t.Fatalf("schema: %v", err)
|
||||
}
|
||||
|
||||
cleanup := func() {
|
||||
_, _ = pool.Exec(ctx, `DELETE FROM alert_rules`)
|
||||
_, _ = pool.Exec(ctx, `DELETE FROM alert_debounce_state`)
|
||||
_, _ = pool.Exec(ctx, `DELETE FROM notification_jobs`)
|
||||
pool.Close()
|
||||
}
|
||||
return NewRuleStore(pool), notify.NewDispatcher(pool), pool, cleanup
|
||||
}
|
||||
|
||||
func gaugeFamily(name string, labels map[string]string, value float64) *dto.MetricFamily {
|
||||
pairs := make([]*dto.LabelPair, 0, len(labels))
|
||||
for k, v := range labels {
|
||||
k, v := k, v
|
||||
pairs = append(pairs, &dto.LabelPair{Name: &k, Value: &v})
|
||||
}
|
||||
n := name
|
||||
return &dto.MetricFamily{
|
||||
Name: &n,
|
||||
Metric: []*dto.Metric{
|
||||
{Label: pairs, Gauge: &dto.Gauge{Value: &value}},
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 1 + 2 / Pruefung 1 + 2: Überschreitung löst eine
|
||||
// Benachrichtigung mit vollständigem Inhalt (Metrik/Tenant/Modul) aus.
|
||||
func TestEvaluate_FiresAlertOnThresholdExceeded(t *testing.T) {
|
||||
rules, dispatcher, pool, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
rule, err := rules.CreateRule(ctx, Rule{
|
||||
MetricName: "nexarch_core_error_rate", Comparison: ComparisonGreaterThan, Threshold: 0.05,
|
||||
Recipient: "ops@acme.example", Description: "Fehlerrate zu hoch",
|
||||
})
|
||||
if err != nil {
|
||||
t.Fatalf("create rule: %v", err)
|
||||
}
|
||||
|
||||
eval := NewEvaluator(rules, dispatcher, pool, time.Hour)
|
||||
families := []*dto.MetricFamily{
|
||||
gaugeFamily("nexarch_core_error_rate", map[string]string{"tenant": "acme", "module": "dms"}, 0.12),
|
||||
}
|
||||
|
||||
fired, err := eval.Evaluate(ctx, families)
|
||||
if err != nil {
|
||||
t.Fatalf("evaluate: %v", err)
|
||||
}
|
||||
if len(fired) != 1 || fired[0].Skipped {
|
||||
t.Fatalf("erwartet genau 1 tatsaechlich ausgeloesten alarm, habe %+v", fired)
|
||||
}
|
||||
if fired[0].RuleID != rule.ID {
|
||||
t.Fatalf("rule id = %q, want %q", fired[0].RuleID, rule.ID)
|
||||
}
|
||||
|
||||
// Pruefung 2: Benachrichtigungsinhalt vollstaendig (Metrik/Tenant/Modul).
|
||||
var payloadJSON []byte
|
||||
if err := pool.QueryRow(ctx, `SELECT payload FROM notification_jobs LIMIT 1`).Scan(&payloadJSON); err != nil {
|
||||
t.Fatalf("notification_jobs lesen: %v", err)
|
||||
}
|
||||
payload := string(payloadJSON)
|
||||
for _, want := range []string{`"metric"`, `nexarch_core_error_rate`, `"tenant"`, `"acme"`, `"module"`, `"dms"`} {
|
||||
if !strings.Contains(payload, want) {
|
||||
t.Errorf("payload enthaelt nicht %q: %s", want, payload)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
func TestEvaluate_DoesNotFireBelowThreshold(t *testing.T) {
|
||||
rules, dispatcher, pool, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
if _, err := rules.CreateRule(ctx, Rule{
|
||||
MetricName: "nexarch_core_error_rate", Comparison: ComparisonGreaterThan, Threshold: 0.05,
|
||||
Recipient: "ops@acme.example",
|
||||
}); err != nil {
|
||||
t.Fatalf("create rule: %v", err)
|
||||
}
|
||||
|
||||
eval := NewEvaluator(rules, dispatcher, pool, time.Hour)
|
||||
families := []*dto.MetricFamily{
|
||||
gaugeFamily("nexarch_core_error_rate", map[string]string{"tenant": "acme"}, 0.01),
|
||||
}
|
||||
|
||||
fired, err := eval.Evaluate(ctx, families)
|
||||
if err != nil {
|
||||
t.Fatalf("evaluate: %v", err)
|
||||
}
|
||||
if len(fired) != 0 {
|
||||
t.Fatalf("erwartet keinen alarm unterhalb des schwellwerts, habe %+v", fired)
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 3 / Pruefung 3: anhaltende Überschreitung erzeugt NICHT
|
||||
// bei jeder Messung eine neue Benachrichtigung.
|
||||
func TestEvaluate_DebouncesRepeatedFiring(t *testing.T) {
|
||||
rules, dispatcher, pool, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
if _, err := rules.CreateRule(ctx, Rule{
|
||||
MetricName: "nexarch_core_error_rate", Comparison: ComparisonGreaterThan, Threshold: 0.05,
|
||||
Recipient: "ops@acme.example",
|
||||
}); err != nil {
|
||||
t.Fatalf("create rule: %v", err)
|
||||
}
|
||||
|
||||
// Langes Debounce-Intervall: der zweite Evaluate-Lauf (simuliert die
|
||||
// naechste Messung bei anhaltend ueberschrittenem Wert) darf keinen
|
||||
// weiteren Job einreihen.
|
||||
eval := NewEvaluator(rules, dispatcher, pool, time.Hour)
|
||||
families := []*dto.MetricFamily{
|
||||
gaugeFamily("nexarch_core_error_rate", map[string]string{"tenant": "acme"}, 0.5),
|
||||
}
|
||||
|
||||
first, err := eval.Evaluate(ctx, families)
|
||||
if err != nil {
|
||||
t.Fatalf("erster evaluate-lauf: %v", err)
|
||||
}
|
||||
if len(first) != 1 || first[0].Skipped {
|
||||
t.Fatalf("erster lauf haette feuern muessen, habe %+v", first)
|
||||
}
|
||||
|
||||
second, err := eval.Evaluate(ctx, families)
|
||||
if err != nil {
|
||||
t.Fatalf("zweiter evaluate-lauf: %v", err)
|
||||
}
|
||||
if len(second) != 1 || !second[0].Skipped {
|
||||
t.Fatalf("zweiter lauf haette gedrosselt werden muessen, habe %+v", second)
|
||||
}
|
||||
|
||||
var count int
|
||||
if err := pool.QueryRow(ctx, `SELECT count(*) FROM notification_jobs`).Scan(&count); err != nil {
|
||||
t.Fatalf("notification_jobs zaehlen: %v", err)
|
||||
}
|
||||
if count != 1 {
|
||||
t.Fatalf("erwartet genau 1 eingereihten job trotz zwei ueberschreitenden messungen, habe %d", count)
|
||||
}
|
||||
}
|
||||
|
||||
// Verschiedene Zeitreihen derselben Regel (unterschiedlicher Tenant) werden
|
||||
// unabhaengig voneinander gedrosselt.
|
||||
func TestEvaluate_DebouncesIndependentlyPerLabelSet(t *testing.T) {
|
||||
rules, dispatcher, pool, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
if _, err := rules.CreateRule(ctx, Rule{
|
||||
MetricName: "nexarch_core_error_rate", Comparison: ComparisonGreaterThan, Threshold: 0.05,
|
||||
Recipient: "ops@acme.example",
|
||||
}); err != nil {
|
||||
t.Fatalf("create rule: %v", err)
|
||||
}
|
||||
|
||||
eval := NewEvaluator(rules, dispatcher, pool, time.Hour)
|
||||
families := []*dto.MetricFamily{
|
||||
{
|
||||
Name: strPtr("nexarch_core_error_rate"),
|
||||
Metric: []*dto.Metric{
|
||||
metricWithLabel("tenant", "acme", 0.5),
|
||||
metricWithLabel("tenant", "beta", 0.6),
|
||||
},
|
||||
},
|
||||
}
|
||||
|
||||
fired, err := eval.Evaluate(ctx, families)
|
||||
if err != nil {
|
||||
t.Fatalf("evaluate: %v", err)
|
||||
}
|
||||
if len(fired) != 2 {
|
||||
t.Fatalf("erwartet 2 unabhaengige alarme (verschiedene tenants), habe %d", len(fired))
|
||||
}
|
||||
for _, a := range fired {
|
||||
if a.Skipped {
|
||||
t.Fatalf("beide tenants sollten beim ersten mal feuern, habe %+v", a)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
func metricWithLabel(name, value string, gaugeValue float64) *dto.Metric {
|
||||
n, v := name, value
|
||||
return &dto.Metric{Label: []*dto.LabelPair{{Name: &n, Value: &v}}, Gauge: &dto.Gauge{Value: &gaugeValue}}
|
||||
}
|
||||
|
||||
func strPtr(s string) *string { return &s }
|
||||
@@ -1,132 +0,0 @@
|
||||
// Package alerting implementiert Core OPS-05: schwellwertbasierte
|
||||
// Alarmierung auf den aus OPS-03 aggregierten Metriken, Zustellung über den
|
||||
// Core-Benachrichtigungs-Dispatcher (CFG-02). Der Alertmanager-Gedanke von
|
||||
// Prometheus/Grafana, aber auf das Nötigste reduziert (Schwellwert, Ziel,
|
||||
// Drosselung) — keine eigene Ausdruckssprache.
|
||||
package alerting
|
||||
|
||||
import (
|
||||
"context"
|
||||
"encoding/json"
|
||||
"errors"
|
||||
"fmt"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
// Comparison legt fest, ob ein Schwellwert nach oben oder unten überwacht
|
||||
// wird — bewusst nur zwei Operatoren, keine eigene Ausdruckssprache
|
||||
// (Ticket-Produkt-DNA).
|
||||
type Comparison string
|
||||
|
||||
const (
|
||||
ComparisonGreaterThan Comparison = "gt"
|
||||
ComparisonLessThan Comparison = "lt"
|
||||
)
|
||||
|
||||
// Rule ist eine Schwellwert-Regel auf einer beliebigen aggregierten Metrik
|
||||
// (Akzeptanzkriterium 1). LabelFilters schränkt optional auf bestimmte
|
||||
// Label-Werte ein (z. B. tenant/module), leer = alle Zeitreihen der Metrik.
|
||||
type Rule struct {
|
||||
ID string
|
||||
MetricName string
|
||||
Comparison Comparison
|
||||
Threshold float64
|
||||
LabelFilters map[string]string
|
||||
Recipient string
|
||||
Description string
|
||||
}
|
||||
|
||||
// RuleStore verwaltet Alert-Regeln in der zentralen Registry-DB.
|
||||
type RuleStore struct {
|
||||
pool *pgxpool.Pool
|
||||
}
|
||||
|
||||
func NewRuleStore(pool *pgxpool.Pool) *RuleStore {
|
||||
return &RuleStore{pool: pool}
|
||||
}
|
||||
|
||||
// CreateRule legt eine neue Schwellwert-Regel an (Akzeptanzkriterium 1).
|
||||
func (s *RuleStore) CreateRule(ctx context.Context, r Rule) (Rule, error) {
|
||||
if r.MetricName == "" || r.Recipient == "" {
|
||||
return Rule{}, errors.New("alerting: metricName und recipient duerfen nicht leer sein")
|
||||
}
|
||||
if r.Comparison != ComparisonGreaterThan && r.Comparison != ComparisonLessThan {
|
||||
return Rule{}, fmt.Errorf("alerting: unbekannter comparison-operator %q", r.Comparison)
|
||||
}
|
||||
if r.LabelFilters == nil {
|
||||
r.LabelFilters = map[string]string{}
|
||||
}
|
||||
filtersJSON, err := json.Marshal(r.LabelFilters)
|
||||
if err != nil {
|
||||
return Rule{}, fmt.Errorf("label-filter serialisieren: %w", err)
|
||||
}
|
||||
|
||||
err = s.pool.QueryRow(ctx, `
|
||||
INSERT INTO alert_rules (metric_name, comparison, threshold, label_filters, recipient, description)
|
||||
VALUES ($1, $2, $3, $4, $5, $6)
|
||||
RETURNING id
|
||||
`, r.MetricName, string(r.Comparison), r.Threshold, filtersJSON, r.Recipient, r.Description).Scan(&r.ID)
|
||||
if err != nil {
|
||||
return Rule{}, fmt.Errorf("regel speichern: %w", err)
|
||||
}
|
||||
return r, nil
|
||||
}
|
||||
|
||||
// ListRules liefert alle konfigurierten Regeln — Grundlage für Evaluate.
|
||||
func (s *RuleStore) ListRules(ctx context.Context) ([]Rule, error) {
|
||||
rows, err := s.pool.Query(ctx, `
|
||||
SELECT id, metric_name, comparison, threshold, label_filters, recipient, description
|
||||
FROM alert_rules ORDER BY created_at
|
||||
`)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("regeln auflisten: %w", err)
|
||||
}
|
||||
defer rows.Close()
|
||||
|
||||
var out []Rule
|
||||
for rows.Next() {
|
||||
var r Rule
|
||||
var comparison string
|
||||
var filtersJSON []byte
|
||||
if err := rows.Scan(&r.ID, &r.MetricName, &comparison, &r.Threshold, &filtersJSON, &r.Recipient, &r.Description); err != nil {
|
||||
return nil, fmt.Errorf("regel lesen: %w", err)
|
||||
}
|
||||
r.Comparison = Comparison(comparison)
|
||||
if err := json.Unmarshal(filtersJSON, &r.LabelFilters); err != nil {
|
||||
return nil, fmt.Errorf("label-filter lesen: %w", err)
|
||||
}
|
||||
out = append(out, r)
|
||||
}
|
||||
return out, rows.Err()
|
||||
}
|
||||
|
||||
// DeleteRule entfernt eine Regel.
|
||||
func (s *RuleStore) DeleteRule(ctx context.Context, id string) error {
|
||||
_, err := s.pool.Exec(ctx, `DELETE FROM alert_rules WHERE id = $1`, id)
|
||||
if err != nil {
|
||||
return fmt.Errorf("regel loeschen: %w", err)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
// debounceStore kapselt die Drosselungs-Zustandstabelle (Akzeptanzkriterium 3).
|
||||
type debounceStore struct {
|
||||
pool *pgxpool.Pool
|
||||
}
|
||||
|
||||
// shouldFire prueft, ob seit dem letzten Alarm fuer ruleKey mindestens
|
||||
// interval vergangen ist — atomar ueber eine bedingte UPDATE/INSERT-
|
||||
// Sequenz, damit zwei gleichzeitige Evaluate-Laeufe (z. B. bei mehreren
|
||||
// Core-Instanzen) nicht beide gleichzeitig alarmieren.
|
||||
func (d *debounceStore) shouldFire(ctx context.Context, ruleKey string, intervalSeconds float64) (bool, error) {
|
||||
tag, err := d.pool.Exec(ctx, `
|
||||
INSERT INTO alert_debounce_state (rule_key, last_fired_at) VALUES ($1, now())
|
||||
ON CONFLICT (rule_key) DO UPDATE SET last_fired_at = now()
|
||||
WHERE alert_debounce_state.last_fired_at <= now() - ($2 * interval '1 second')
|
||||
`, ruleKey, intervalSeconds)
|
||||
if err != nil {
|
||||
return false, fmt.Errorf("drosselungszustand pruefen: %w", err)
|
||||
}
|
||||
return tag.RowsAffected() == 1, nil
|
||||
}
|
||||
@@ -0,0 +1,20 @@
|
||||
package apiserver
|
||||
|
||||
import "context"
|
||||
|
||||
type contextKey int
|
||||
|
||||
const requestContextKey contextKey = iota
|
||||
|
||||
// RequestContext ist der Tenant-/Benutzerkontext, den die Middleware-Kette
|
||||
// fuer nachgelagerte Handler bereitstellt (Akzeptanzkriterium 3).
|
||||
type RequestContext struct {
|
||||
UserID string
|
||||
TenantSlug string
|
||||
}
|
||||
|
||||
// FromContext liest den von der Middleware gesetzten Kontext.
|
||||
func FromContext(ctx context.Context) (RequestContext, bool) {
|
||||
rc, ok := ctx.Value(requestContextKey).(RequestContext)
|
||||
return rc, ok
|
||||
}
|
||||
@@ -0,0 +1,31 @@
|
||||
// Package apiserver implementiert Core API-01: das REST-Grundgerüst mit
|
||||
// URL-Versionierung, einheitlichem Fehlerformat und Middleware-Kette
|
||||
// (Auth, Tenant-/Benutzerkontext, Logging).
|
||||
package apiserver
|
||||
|
||||
import (
|
||||
"encoding/json"
|
||||
"net/http"
|
||||
)
|
||||
|
||||
// errorBody ist das EINE Fehlerschema fuer alle Endpunkte unter /api/{version}/
|
||||
// (Akzeptanzkriterium 2).
|
||||
type errorBody struct {
|
||||
Error struct {
|
||||
Code string `json:"code"`
|
||||
Message string `json:"message"`
|
||||
} `json:"error"`
|
||||
}
|
||||
|
||||
// WriteError schreibt einen Fehler im einheitlichen Schema. code ist ein
|
||||
// stabiler, maschinenlesbarer Bezeichner (z.B. "unauthenticated"), message
|
||||
// ein fuer Menschen lesbarer deutscher Text.
|
||||
func WriteError(w http.ResponseWriter, status int, code, message string) {
|
||||
var body errorBody
|
||||
body.Error.Code = code
|
||||
body.Error.Message = message
|
||||
|
||||
w.Header().Set("Content-Type", "application/json")
|
||||
w.WriteHeader(status)
|
||||
_ = json.NewEncoder(w).Encode(body)
|
||||
}
|
||||
@@ -0,0 +1,56 @@
|
||||
package apiserver
|
||||
|
||||
import (
|
||||
"context"
|
||||
"log/slog"
|
||||
"net/http"
|
||||
"time"
|
||||
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/auth"
|
||||
)
|
||||
|
||||
// authAndTenantContext prueft die Session (wiederverwendet auth.TokenIssuer.Verify
|
||||
// aus IAM-02 — keine zweite JWT-Implementierung) und setzt bei Erfolg
|
||||
// RequestContext fuer nachgelagerte Handler (Akzeptanzkriterium 3). Anders
|
||||
// als auth.RequireAuth (Klartext-Fehler) antwortet diese Middleware im
|
||||
// einheitlichen API-01-Fehlerschema (Akzeptanzkriterium 2), damit ALLE
|
||||
// Endpunkte unter /api/{version}/ dasselbe Format liefern, auch bei
|
||||
// Auth-Fehlern.
|
||||
func authAndTenantContext(issuer *auth.TokenIssuer, next http.HandlerFunc) http.HandlerFunc {
|
||||
return func(w http.ResponseWriter, r *http.Request) {
|
||||
cookie, err := r.Cookie(auth.CookieName)
|
||||
if err != nil {
|
||||
WriteError(w, http.StatusUnauthorized, "unauthenticated", "nicht angemeldet")
|
||||
return
|
||||
}
|
||||
|
||||
claims, err := issuer.Verify(cookie.Value)
|
||||
if err != nil {
|
||||
WriteError(w, http.StatusUnauthorized, "unauthenticated", "nicht angemeldet")
|
||||
return
|
||||
}
|
||||
|
||||
rc := RequestContext{UserID: claims.UserID, TenantSlug: claims.TenantSlug}
|
||||
next(w, r.WithContext(context.WithValue(r.Context(), requestContextKey, rc)))
|
||||
}
|
||||
}
|
||||
|
||||
type statusRecorder struct {
|
||||
http.ResponseWriter
|
||||
status int
|
||||
}
|
||||
|
||||
func (s *statusRecorder) WriteHeader(code int) {
|
||||
s.status = code
|
||||
s.ResponseWriter.WriteHeader(code)
|
||||
}
|
||||
|
||||
// loggingMiddleware protokolliert jede Anfrage strukturiert.
|
||||
func loggingMiddleware(next http.HandlerFunc) http.HandlerFunc {
|
||||
return func(w http.ResponseWriter, r *http.Request) {
|
||||
rec := &statusRecorder{ResponseWriter: w, status: http.StatusOK}
|
||||
start := time.Now()
|
||||
next(rec, r)
|
||||
slog.Info("api-anfrage", "method", r.Method, "path", r.URL.Path, "status", rec.status, "dauer", time.Since(start))
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,37 @@
|
||||
package apiserver
|
||||
|
||||
import (
|
||||
"net/http"
|
||||
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/auth"
|
||||
)
|
||||
|
||||
// Server registriert versionierte API-Routen (Akzeptanzkriterium 1: unter
|
||||
// /api/{version}/...) und verdrahtet fuer jede Route dieselbe Middleware-
|
||||
// Kette (Logging -> Auth+Tenantkontext -> Handler).
|
||||
type Server struct {
|
||||
mux *http.ServeMux
|
||||
issuer *auth.TokenIssuer
|
||||
}
|
||||
|
||||
func NewServer(issuer *auth.TokenIssuer) *Server {
|
||||
return &Server{mux: http.NewServeMux(), issuer: issuer}
|
||||
}
|
||||
|
||||
// Handle registriert pattern unter der angegebenen Version, z.B.
|
||||
// Handle("v1", "/things", h) -> erreichbar unter /api/v1/things. Verschiedene
|
||||
// Versionen sind unabhaengige Pfade — eine neue Version beeintraechtigt
|
||||
// bestehende nicht (Akzeptanzkriterium 1 / Pruefung 3).
|
||||
func (s *Server) Handle(version, pattern string, h http.HandlerFunc) {
|
||||
full := "/api/" + version + pattern
|
||||
s.mux.HandleFunc(full, loggingMiddleware(authAndTenantContext(s.issuer, h)))
|
||||
}
|
||||
|
||||
// HandleV1 ist die Kurzform fuer die aktuelle Hauptversion.
|
||||
func (s *Server) HandleV1(pattern string, h http.HandlerFunc) {
|
||||
s.Handle("v1", pattern, h)
|
||||
}
|
||||
|
||||
func (s *Server) Handler() http.Handler {
|
||||
return s.mux
|
||||
}
|
||||
@@ -0,0 +1,149 @@
|
||||
package apiserver
|
||||
|
||||
import (
|
||||
"encoding/json"
|
||||
"net/http"
|
||||
"net/http/httptest"
|
||||
"testing"
|
||||
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/auth"
|
||||
)
|
||||
|
||||
func newTestServer() (*Server, *auth.TokenIssuer) {
|
||||
issuer := auth.NewTokenIssuer("test-secret-nur-fuer-tests")
|
||||
return NewServer(issuer), issuer
|
||||
}
|
||||
|
||||
func withAuthCookie(req *http.Request, token string) *http.Request {
|
||||
req.AddCookie(&http.Cookie{Name: auth.CookieName, Value: token})
|
||||
return req
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 1: API unter versioniertem Pfad erreichbar.
|
||||
func TestHandleV1_RegistersUnderVersionedPath(t *testing.T) {
|
||||
srv, issuer := newTestServer()
|
||||
srv.HandleV1("/things", func(w http.ResponseWriter, r *http.Request) {
|
||||
w.WriteHeader(http.StatusOK)
|
||||
})
|
||||
|
||||
token, err := issuer.Issue("user-1", "acme")
|
||||
if err != nil {
|
||||
t.Fatalf("issue: %v", err)
|
||||
}
|
||||
|
||||
req := withAuthCookie(httptest.NewRequest(http.MethodGet, "/api/v1/things", nil), token)
|
||||
rec := httptest.NewRecorder()
|
||||
srv.Handler().ServeHTTP(rec, req)
|
||||
if rec.Code != http.StatusOK {
|
||||
t.Fatalf("status = %d, want 200", rec.Code)
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 + Pruefung 1 (Stichprobe): mehrere Endpunkte liefern
|
||||
// bei fehlerhafter Anfrage dasselbe Fehlerschema.
|
||||
func TestErrorFormat_ConsistentAcrossEndpoints(t *testing.T) {
|
||||
srv, _ := newTestServer()
|
||||
srv.HandleV1("/endpunkt-a", func(w http.ResponseWriter, r *http.Request) { w.WriteHeader(http.StatusOK) })
|
||||
srv.HandleV1("/endpunkt-b", func(w http.ResponseWriter, r *http.Request) { w.WriteHeader(http.StatusOK) })
|
||||
|
||||
for _, path := range []string{"/api/v1/endpunkt-a", "/api/v1/endpunkt-b"} {
|
||||
req := httptest.NewRequest(http.MethodGet, path, nil) // ohne cookie -> 401
|
||||
rec := httptest.NewRecorder()
|
||||
srv.Handler().ServeHTTP(rec, req)
|
||||
|
||||
if rec.Code != http.StatusUnauthorized {
|
||||
t.Fatalf("%s: status = %d, want 401", path, rec.Code)
|
||||
}
|
||||
var body errorBody
|
||||
if err := json.Unmarshal(rec.Body.Bytes(), &body); err != nil {
|
||||
t.Fatalf("%s: fehlerantwort nicht im erwarteten json-schema: %v (body: %s)", path, err, rec.Body.String())
|
||||
}
|
||||
if body.Error.Code == "" || body.Error.Message == "" {
|
||||
t.Fatalf("%s: erwartet nicht-leeren code/message, habe %+v", path, body)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 3 + Pruefung 2: Middleware-Kette setzt Tenant-/
|
||||
// Benutzerkontext zuverlaessig, nachweislich fuer mehrere Endpunkte.
|
||||
func TestMiddleware_SetsRequestContextForEveryEndpoint(t *testing.T) {
|
||||
srv, issuer := newTestServer()
|
||||
|
||||
var gotA, gotB RequestContext
|
||||
srv.HandleV1("/kontext-a", func(w http.ResponseWriter, r *http.Request) {
|
||||
gotA, _ = FromContext(r.Context())
|
||||
w.WriteHeader(http.StatusOK)
|
||||
})
|
||||
srv.HandleV1("/kontext-b", func(w http.ResponseWriter, r *http.Request) {
|
||||
gotB, _ = FromContext(r.Context())
|
||||
w.WriteHeader(http.StatusOK)
|
||||
})
|
||||
|
||||
token, err := issuer.Issue("user-42", "tenant-x")
|
||||
if err != nil {
|
||||
t.Fatalf("issue: %v", err)
|
||||
}
|
||||
|
||||
for path, got := range map[string]*RequestContext{"/api/v1/kontext-a": &gotA, "/api/v1/kontext-b": &gotB} {
|
||||
req := withAuthCookie(httptest.NewRequest(http.MethodGet, path, nil), token)
|
||||
rec := httptest.NewRecorder()
|
||||
srv.Handler().ServeHTTP(rec, req)
|
||||
if rec.Code != http.StatusOK {
|
||||
t.Fatalf("%s: status = %d, want 200", path, rec.Code)
|
||||
}
|
||||
if got.UserID != "user-42" || got.TenantSlug != "tenant-x" {
|
||||
t.Fatalf("%s: request-context unerwartet: %+v", path, *got)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 1 + Pruefung 3: eine neue v2-Route laesst sich anlegen,
|
||||
// ohne v1 zu beeintraechtigen.
|
||||
func TestVersioning_V2DoesNotAffectV1(t *testing.T) {
|
||||
srv, issuer := newTestServer()
|
||||
srv.HandleV1("/things", func(w http.ResponseWriter, r *http.Request) {
|
||||
w.Write([]byte("v1-antwort"))
|
||||
})
|
||||
srv.Handle("v2", "/things", func(w http.ResponseWriter, r *http.Request) {
|
||||
w.Write([]byte("v2-antwort"))
|
||||
})
|
||||
|
||||
token, err := issuer.Issue("user-1", "acme")
|
||||
if err != nil {
|
||||
t.Fatalf("issue: %v", err)
|
||||
}
|
||||
|
||||
reqV1 := withAuthCookie(httptest.NewRequest(http.MethodGet, "/api/v1/things", nil), token)
|
||||
recV1 := httptest.NewRecorder()
|
||||
srv.Handler().ServeHTTP(recV1, reqV1)
|
||||
if recV1.Body.String() != "v1-antwort" {
|
||||
t.Fatalf("v1 antwort = %q, want v1-antwort", recV1.Body.String())
|
||||
}
|
||||
|
||||
reqV2 := withAuthCookie(httptest.NewRequest(http.MethodGet, "/api/v2/things", nil), token)
|
||||
recV2 := httptest.NewRecorder()
|
||||
srv.Handler().ServeHTTP(recV2, reqV2)
|
||||
if recV2.Body.String() != "v2-antwort" {
|
||||
t.Fatalf("v2 antwort = %q, want v2-antwort", recV2.Body.String())
|
||||
}
|
||||
|
||||
// v1 nach dem Anlegen von v2 erneut pruefen — unveraendert.
|
||||
reqV1Again := withAuthCookie(httptest.NewRequest(http.MethodGet, "/api/v1/things", nil), token)
|
||||
recV1Again := httptest.NewRecorder()
|
||||
srv.Handler().ServeHTTP(recV1Again, reqV1Again)
|
||||
if recV1Again.Body.String() != "v1-antwort" {
|
||||
t.Fatalf("v1 antwort nach v2-anlage = %q, want weiterhin v1-antwort", recV1Again.Body.String())
|
||||
}
|
||||
}
|
||||
|
||||
func TestAuthAndTenantContext_RejectsInvalidToken(t *testing.T) {
|
||||
srv, _ := newTestServer()
|
||||
srv.HandleV1("/geschuetzt", func(w http.ResponseWriter, r *http.Request) { w.WriteHeader(http.StatusOK) })
|
||||
|
||||
req := withAuthCookie(httptest.NewRequest(http.MethodGet, "/api/v1/geschuetzt", nil), "kaputtes.token.hier")
|
||||
rec := httptest.NewRecorder()
|
||||
srv.Handler().ServeHTTP(rec, req)
|
||||
if rec.Code != http.StatusUnauthorized {
|
||||
t.Fatalf("status = %d, want 401", rec.Code)
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,67 @@
|
||||
package auth
|
||||
|
||||
import (
|
||||
"encoding/json"
|
||||
"net/http"
|
||||
"time"
|
||||
)
|
||||
|
||||
// Handler stellt Login/Logout als HTTP-Endpunkte bereit. Registrierung,
|
||||
// Passwort-Reset, 2FA, SSO/LDAP und Rate-Limiting sind ausdruecklich nicht
|
||||
// Teil dieser Kachel (siehe IAM-03..07).
|
||||
type Handler struct {
|
||||
login *LoginService
|
||||
}
|
||||
|
||||
func NewHandler(login *LoginService) *Handler {
|
||||
return &Handler{login: login}
|
||||
}
|
||||
|
||||
type loginRequest struct {
|
||||
Email string `json:"email"`
|
||||
Password string `json:"password"`
|
||||
}
|
||||
|
||||
func (h *Handler) Login(w http.ResponseWriter, r *http.Request) {
|
||||
var req loginRequest
|
||||
if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
|
||||
http.Error(w, "ungueltige Anfrage", http.StatusBadRequest)
|
||||
return
|
||||
}
|
||||
|
||||
token, err := h.login.Login(r.Context(), req.Email, req.Password)
|
||||
if err != nil {
|
||||
http.Error(w, ErrInvalidCredentials.Error(), http.StatusUnauthorized)
|
||||
return
|
||||
}
|
||||
|
||||
http.SetCookie(w, &http.Cookie{
|
||||
Name: CookieName,
|
||||
Value: token,
|
||||
Path: "/",
|
||||
HttpOnly: true,
|
||||
Secure: true,
|
||||
SameSite: http.SameSiteStrictMode,
|
||||
MaxAge: int(AccessTokenTTL.Seconds()),
|
||||
})
|
||||
w.WriteHeader(http.StatusOK)
|
||||
}
|
||||
|
||||
// Logout loescht das Session-Cookie. Da JWT hier bewusst zustandslos bleibt
|
||||
// (kein serverseitiger Blocklist-Speicher — das waere ueber "Grundgerüst"
|
||||
// hinaus und widerspraeche der projektweiten zustandslosen-JWT-Entscheidung),
|
||||
// bleibt ein bereits ausgestelltes Token bis zu seinem Ablauf technisch
|
||||
// gueltig, wenn es separat vom Cookie extrahiert und wiederverwendet wird.
|
||||
func (h *Handler) Logout(w http.ResponseWriter, r *http.Request) {
|
||||
http.SetCookie(w, &http.Cookie{
|
||||
Name: CookieName,
|
||||
Value: "",
|
||||
Path: "/",
|
||||
HttpOnly: true,
|
||||
Secure: true,
|
||||
SameSite: http.SameSiteStrictMode,
|
||||
MaxAge: -1,
|
||||
Expires: time.Unix(0, 0),
|
||||
})
|
||||
w.WriteHeader(http.StatusOK)
|
||||
}
|
||||
@@ -0,0 +1,56 @@
|
||||
package auth
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/user"
|
||||
)
|
||||
|
||||
var ErrInvalidCredentials = errors.New("auth: E-Mail oder Passwort falsch")
|
||||
|
||||
// LoginService arbeitet gegen GENAU EINE Tenant-Datenbank (uebergeben ueber
|
||||
// den TenantUserStore-Pool) — das Login ist damit strukturell auf den
|
||||
// richtigen Tenant gescopt, siehe user.TenantUserStore.GetByEmailForAuth.
|
||||
type LoginService struct {
|
||||
users *user.TenantUserStore
|
||||
issuer *TokenIssuer
|
||||
// tenantSlug identifiziert im ausgestellten Token, gegen welchen Mandanten
|
||||
// eingeloggt wurde (fuer nachgelagerte Pruefungen, z.B. Middleware-Logs).
|
||||
tenantSlug string
|
||||
}
|
||||
|
||||
func NewLoginService(users *user.TenantUserStore, issuer *TokenIssuer, tenantSlug string) *LoginService {
|
||||
return &LoginService{users: users, issuer: issuer, tenantSlug: tenantSlug}
|
||||
}
|
||||
|
||||
// Login liefert bei falscher E-Mail UND bei falschem Passwort denselben
|
||||
// Fehler (ErrInvalidCredentials), um keine Rueckschluesse auf die Existenz
|
||||
// eines Kontos zuzulassen (User-Enumeration-Schutz).
|
||||
func (s *LoginService) Login(ctx context.Context, email, password string) (string, error) {
|
||||
creds, err := s.users.GetByEmailForAuth(ctx, email)
|
||||
if err != nil {
|
||||
// Trotzdem einen bcrypt-Vergleich gegen einen Dummy-Hash ausfuehren,
|
||||
// damit die Antwortzeit bei unbekannter E-Mail nicht messbar kuerzer
|
||||
// ist als bei falschem Passwort (Timing-Seitenkanal).
|
||||
VerifyPassword(dummyHash, password)
|
||||
return "", ErrInvalidCredentials
|
||||
}
|
||||
|
||||
if creds.User.Status != user.StatusActive {
|
||||
return "", ErrInvalidCredentials
|
||||
}
|
||||
|
||||
if !VerifyPassword(creds.PasswordHash, password) {
|
||||
return "", ErrInvalidCredentials
|
||||
}
|
||||
|
||||
return s.issuer.Issue(creds.User.ID, s.tenantSlug)
|
||||
}
|
||||
|
||||
// dummyHash ist ein echter bcrypt-Hash (Kostenfaktor BcryptCost) eines
|
||||
// beliebigen Platzhalter-Klartexts — bewusst KEIN kaputtes Format, da
|
||||
// bcrypt.CompareHashAndPassword bei ungueltigem Hash sofort ohne den
|
||||
// eigentlichen Kostenfaktor-Vergleich zurueckkehrt und die
|
||||
// Timing-Angleichung damit wirkungslos waere.
|
||||
const dummyHash = "$2a$12$cmwiETrG9DK5/uTM2fg4uetngYUspKjME5P8fNpk0QYTaO64N0r3C"
|
||||
@@ -0,0 +1,177 @@
|
||||
package auth
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"fmt"
|
||||
"net/http"
|
||||
"net/http/httptest"
|
||||
"os"
|
||||
"strings"
|
||||
"testing"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/user"
|
||||
)
|
||||
|
||||
const usersSchema = `
|
||||
CREATE EXTENSION IF NOT EXISTS pgcrypto;
|
||||
CREATE TABLE users (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
email TEXT NOT NULL UNIQUE,
|
||||
name TEXT NOT NULL,
|
||||
status TEXT NOT NULL DEFAULT 'active',
|
||||
password_hash TEXT NOT NULL DEFAULT '',
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);`
|
||||
|
||||
func setupTenantDB(t *testing.T, dbName string) *pgxpool.Pool {
|
||||
t.Helper()
|
||||
adminDSN := os.Getenv("TEST_ADMIN_DSN")
|
||||
if adminDSN == "" {
|
||||
t.Skip("TEST_ADMIN_DSN nicht gesetzt, Integrationstest uebersprungen")
|
||||
}
|
||||
ctx := context.Background()
|
||||
|
||||
adminPool, err := pgxpool.New(ctx, adminDSN)
|
||||
if err != nil {
|
||||
t.Fatalf("admin pool: %v", err)
|
||||
}
|
||||
_, _ = adminPool.Exec(ctx, fmt.Sprintf(`DROP DATABASE IF EXISTS %q`, dbName))
|
||||
if _, err := adminPool.Exec(ctx, fmt.Sprintf(`CREATE DATABASE %q`, dbName)); err != nil {
|
||||
t.Fatalf("testdatenbank anlegen: %v", err)
|
||||
}
|
||||
|
||||
dsn := strings.Replace(adminDSN, "/postgres?", "/"+dbName+"?", 1)
|
||||
pool, err := pgxpool.New(ctx, dsn)
|
||||
if err != nil {
|
||||
t.Fatalf("connect testdatenbank: %v", err)
|
||||
}
|
||||
if _, err := pool.Exec(ctx, usersSchema); err != nil {
|
||||
t.Fatalf("schema anwenden: %v", err)
|
||||
}
|
||||
|
||||
t.Cleanup(func() {
|
||||
pool.Close()
|
||||
_, _ = adminPool.Exec(ctx, fmt.Sprintf(`DROP DATABASE IF EXISTS %q`, dbName))
|
||||
adminPool.Close()
|
||||
})
|
||||
return pool
|
||||
}
|
||||
|
||||
func createUserWithPassword(t *testing.T, store *user.TenantUserStore, email, password string) user.User {
|
||||
t.Helper()
|
||||
ctx := context.Background()
|
||||
u, err := store.Create(ctx, email, "Test User")
|
||||
if err != nil {
|
||||
t.Fatalf("create user: %v", err)
|
||||
}
|
||||
hash, err := HashPassword(password)
|
||||
if err != nil {
|
||||
t.Fatalf("hash password: %v", err)
|
||||
}
|
||||
if err := store.SetPasswordHash(ctx, u.ID, hash); err != nil {
|
||||
t.Fatalf("set password: %v", err)
|
||||
}
|
||||
return u
|
||||
}
|
||||
|
||||
func TestLoginService_SuccessAndWrongPassword(t *testing.T) {
|
||||
pool := setupTenantDB(t, "test_iam02_login")
|
||||
store := user.NewTenantUserStore(pool)
|
||||
createUserWithPassword(t, store, "alice@example.com", "korrektes-passwort")
|
||||
|
||||
issuer := NewTokenIssuer("test-secret-nur-fuer-tests")
|
||||
login := NewLoginService(store, issuer, "acme")
|
||||
|
||||
token, err := login.Login(context.Background(), "alice@example.com", "korrektes-passwort")
|
||||
if err != nil {
|
||||
t.Fatalf("login: %v", err)
|
||||
}
|
||||
if token == "" {
|
||||
t.Fatal("erwartet nicht-leeres token")
|
||||
}
|
||||
|
||||
if _, err := login.Login(context.Background(), "alice@example.com", "falsches-passwort"); !errors.Is(err, ErrInvalidCredentials) {
|
||||
t.Fatalf("erwartet ErrInvalidCredentials, habe %v", err)
|
||||
}
|
||||
|
||||
if _, err := login.Login(context.Background(), "unbekannt@example.com", "irgendwas"); !errors.Is(err, ErrInvalidCredentials) {
|
||||
t.Fatalf("erwartet ErrInvalidCredentials bei unbekannter email, habe %v", err)
|
||||
}
|
||||
}
|
||||
|
||||
// Pruefung 1: kein Cross-Tenant-Login moeglich, obwohl dieselbe E-Mail in
|
||||
// zwei unterschiedlichen Tenant-Datenbanken mit unterschiedlichen Passwoertern
|
||||
// existiert.
|
||||
func TestLoginService_NoCrossTenantLogin(t *testing.T) {
|
||||
poolA := setupTenantDB(t, "test_iam02_tenant_a")
|
||||
poolB := setupTenantDB(t, "test_iam02_tenant_b")
|
||||
|
||||
storeA := user.NewTenantUserStore(poolA)
|
||||
storeB := user.NewTenantUserStore(poolB)
|
||||
createUserWithPassword(t, storeA, "shared@example.com", "passwort-tenant-a")
|
||||
createUserWithPassword(t, storeB, "shared@example.com", "passwort-tenant-b")
|
||||
|
||||
issuer := NewTokenIssuer("test-secret-nur-fuer-tests")
|
||||
loginA := NewLoginService(storeA, issuer, "tenant-a")
|
||||
|
||||
// Login gegen Tenant A mit dem Passwort von Tenant B darf nicht klappen,
|
||||
// obwohl die E-Mail-Adresse identisch ist — die Store-Instanz kennt
|
||||
// strukturell nur die Zeilen ihrer eigenen Datenbank.
|
||||
if _, err := loginA.Login(context.Background(), "shared@example.com", "passwort-tenant-b"); !errors.Is(err, ErrInvalidCredentials) {
|
||||
t.Fatalf("erwartet ErrInvalidCredentials fuer fremdes tenant-passwort, habe %v", err)
|
||||
}
|
||||
|
||||
token, err := loginA.Login(context.Background(), "shared@example.com", "passwort-tenant-a")
|
||||
if err != nil {
|
||||
t.Fatalf("login gegen eigenen tenant sollte klappen: %v", err)
|
||||
}
|
||||
claims, err := issuer.Verify(token)
|
||||
if err != nil {
|
||||
t.Fatalf("verify: %v", err)
|
||||
}
|
||||
if claims.TenantSlug != "tenant-a" {
|
||||
t.Fatalf("token tenant = %q, want tenant-a", claims.TenantSlug)
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 3: geschuetzte Route ohne gueltige Session nicht erreichbar.
|
||||
func TestRequireAuth_BlocksWithoutValidCookie(t *testing.T) {
|
||||
issuer := NewTokenIssuer("test-secret-nur-fuer-tests")
|
||||
protected := RequireAuth(issuer, func(w http.ResponseWriter, r *http.Request) {
|
||||
w.WriteHeader(http.StatusOK)
|
||||
})
|
||||
|
||||
// Kein Cookie.
|
||||
req := httptest.NewRequest(http.MethodGet, "/geschuetzt", nil)
|
||||
rec := httptest.NewRecorder()
|
||||
protected(rec, req)
|
||||
if rec.Code != http.StatusUnauthorized {
|
||||
t.Fatalf("ohne cookie: status = %d, want 401", rec.Code)
|
||||
}
|
||||
|
||||
// Manipuliertes Cookie.
|
||||
req = httptest.NewRequest(http.MethodGet, "/geschuetzt", nil)
|
||||
req.AddCookie(&http.Cookie{Name: CookieName, Value: "kaputt.token.hier"})
|
||||
rec = httptest.NewRecorder()
|
||||
protected(rec, req)
|
||||
if rec.Code != http.StatusUnauthorized {
|
||||
t.Fatalf("mit kaputtem cookie: status = %d, want 401", rec.Code)
|
||||
}
|
||||
|
||||
// Gueltiges Token.
|
||||
token, err := issuer.Issue("user-1", "acme")
|
||||
if err != nil {
|
||||
t.Fatalf("issue: %v", err)
|
||||
}
|
||||
req = httptest.NewRequest(http.MethodGet, "/geschuetzt", nil)
|
||||
req.AddCookie(&http.Cookie{Name: CookieName, Value: token})
|
||||
rec = httptest.NewRecorder()
|
||||
protected(rec, req)
|
||||
if rec.Code != http.StatusOK {
|
||||
t.Fatalf("mit gueltigem cookie: status = %d, want 200", rec.Code)
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,41 @@
|
||||
package auth
|
||||
|
||||
import (
|
||||
"context"
|
||||
"net/http"
|
||||
)
|
||||
|
||||
const CookieName = "nexarch_session"
|
||||
|
||||
type contextKey int
|
||||
|
||||
const claimsContextKey contextKey = iota
|
||||
|
||||
// RequireAuth schuetzt eine Route: ohne gueltiges, nicht abgelaufenes Token
|
||||
// im Session-Cookie wird 401 zurueckgegeben und der Handler nicht aufgerufen
|
||||
// (IAM-02 Akzeptanzkriterium 3).
|
||||
func RequireAuth(issuer *TokenIssuer, next http.HandlerFunc) http.HandlerFunc {
|
||||
return func(w http.ResponseWriter, r *http.Request) {
|
||||
cookie, err := r.Cookie(CookieName)
|
||||
if err != nil {
|
||||
http.Error(w, "nicht angemeldet", http.StatusUnauthorized)
|
||||
return
|
||||
}
|
||||
|
||||
claims, err := issuer.Verify(cookie.Value)
|
||||
if err != nil {
|
||||
http.Error(w, "nicht angemeldet", http.StatusUnauthorized)
|
||||
return
|
||||
}
|
||||
|
||||
ctx := context.WithValue(r.Context(), claimsContextKey, claims)
|
||||
next(w, r.WithContext(ctx))
|
||||
}
|
||||
}
|
||||
|
||||
// ClaimsFromContext liest die Claims, die RequireAuth in den Request-Context
|
||||
// gelegt hat.
|
||||
func ClaimsFromContext(ctx context.Context) (*Claims, bool) {
|
||||
c, ok := ctx.Value(claimsContextKey).(*Claims)
|
||||
return c, ok
|
||||
}
|
||||
@@ -0,0 +1,28 @@
|
||||
// Package auth implementiert Core IAM-02: Login/Logout, Passwort-Hashing und
|
||||
// die Middleware zum Schutz von Routen. Autorisierung (was ein Benutzer darf)
|
||||
// ist ausdruecklich NICHT Teil dieses Pakets, siehe RBAC-01 — auth prueft nur
|
||||
// "wer bin ich" (Casbin-Architekturprinzip, siehe IAM-02-Ticket).
|
||||
package auth
|
||||
|
||||
import "golang.org/x/crypto/bcrypt"
|
||||
|
||||
// BcryptCost ist bewusst explizit festgelegt statt bcrypt.DefaultCost (10)
|
||||
// unreflektiert zu uebernehmen (IAM-02 Akzeptanzkriterium 4). Kostenfaktor 12
|
||||
// wurde gegen die Ziel-Login-Latenz benchmarkt, siehe password_bench_test.go
|
||||
// und den Pruefungs-Eintrag in der Commit-Nachricht.
|
||||
const BcryptCost = 12
|
||||
|
||||
func HashPassword(plain string) (string, error) {
|
||||
hash, err := bcrypt.GenerateFromPassword([]byte(plain), BcryptCost)
|
||||
if err != nil {
|
||||
return "", err
|
||||
}
|
||||
return string(hash), nil
|
||||
}
|
||||
|
||||
// VerifyPassword ist timing-safe: bcrypt.CompareHashAndPassword vergleicht
|
||||
// konstant in der Zeit bzgl. des Hash-Inhalts (Referenzimplementierung fuer
|
||||
// die projektweite Timing-safe-Vergleich-Konvention aus IAM-02).
|
||||
func VerifyPassword(hash, plain string) bool {
|
||||
return bcrypt.CompareHashAndPassword([]byte(hash), []byte(plain)) == nil
|
||||
}
|
||||
@@ -0,0 +1,42 @@
|
||||
package auth
|
||||
|
||||
import (
|
||||
"testing"
|
||||
"time"
|
||||
)
|
||||
|
||||
// TargetLoginLatency ist der Zielwert aus IAM-02 Akzeptanzkriterium 4: der
|
||||
// bcrypt-Vergleich allein darf die Login-Latenz nicht dominieren. 400ms ist
|
||||
// grosszuegig genug, um auf unterschiedlicher Hardware stabil zu sein, aber
|
||||
// eng genug, um eine versehentliche Kostenfaktor-Explosion (z.B. 16 statt 12)
|
||||
// zuverlaessig aufzudecken.
|
||||
const TargetLoginLatency = 400 * time.Millisecond
|
||||
|
||||
// TestBcryptCostAgainstLatencyTarget misst die tatsaechliche Dauer eines
|
||||
// Passwort-Vergleichs mit dem festgelegten BcryptCost und dokumentiert das
|
||||
// Ergebnis (IAM-02 Pruefung 4).
|
||||
func TestBcryptCostAgainstLatencyTarget(t *testing.T) {
|
||||
hash, err := HashPassword("benchmark-passwort")
|
||||
if err != nil {
|
||||
t.Fatalf("hash: %v", err)
|
||||
}
|
||||
|
||||
start := time.Now()
|
||||
if !VerifyPassword(hash, "benchmark-passwort") {
|
||||
t.Fatal("verifikation haette erfolgreich sein muessen")
|
||||
}
|
||||
elapsed := time.Since(start)
|
||||
|
||||
t.Logf("bcrypt-vergleich mit cost=%d dauerte %s (ziel: unter %s)", BcryptCost, elapsed, TargetLoginLatency)
|
||||
if elapsed > TargetLoginLatency {
|
||||
t.Fatalf("bcrypt-vergleich zu langsam: %s > ziel %s", elapsed, TargetLoginLatency)
|
||||
}
|
||||
}
|
||||
|
||||
func BenchmarkVerifyPassword(b *testing.B) {
|
||||
hash, _ := HashPassword("benchmark-passwort")
|
||||
b.ResetTimer()
|
||||
for i := 0; i < b.N; i++ {
|
||||
VerifyPassword(hash, "benchmark-passwort")
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,28 @@
|
||||
package auth
|
||||
|
||||
import "testing"
|
||||
|
||||
func TestHashAndVerifyPassword(t *testing.T) {
|
||||
hash, err := HashPassword("s3hr-geheim!")
|
||||
if err != nil {
|
||||
t.Fatalf("hash: %v", err)
|
||||
}
|
||||
if hash == "s3hr-geheim!" {
|
||||
t.Fatal("passwort wurde nicht gehasht")
|
||||
}
|
||||
if !VerifyPassword(hash, "s3hr-geheim!") {
|
||||
t.Fatal("erwartet erfolgreiche verifikation")
|
||||
}
|
||||
if VerifyPassword(hash, "falsches-passwort") {
|
||||
t.Fatal("erwartet fehlgeschlagene verifikation")
|
||||
}
|
||||
}
|
||||
|
||||
func TestDummyHashIsValidBcryptHash(t *testing.T) {
|
||||
// Stellt sicher, dass der Timing-Angleichs-Hash in login.go tatsaechlich
|
||||
// ein gueltiges bcrypt-Format hat und den vollen Kostenfaktor durchlaeuft
|
||||
// (siehe Kommentar dort) statt sofort mit einem Format-Fehler abzubrechen.
|
||||
if VerifyPassword(dummyHash, "irgendein-text") {
|
||||
t.Fatal("dummyHash sollte fuer beliebigen text nicht passen")
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,63 @@
|
||||
package auth
|
||||
|
||||
import (
|
||||
"errors"
|
||||
"time"
|
||||
|
||||
"github.com/golang-jwt/jwt/v5"
|
||||
)
|
||||
|
||||
// AccessTokenTTL ist bewusst kurz gehalten (Session-Ablauf statt langlebiger
|
||||
// Tokens), passend zur "so vertrauenswuerdig wie noetig"-Produkt-DNA.
|
||||
const AccessTokenTTL = 30 * time.Minute
|
||||
|
||||
var ErrInvalidToken = errors.New("auth: ungueltiges oder abgelaufenes token")
|
||||
|
||||
type Claims struct {
|
||||
UserID string `json:"uid"`
|
||||
TenantSlug string `json:"tenant"`
|
||||
jwt.RegisteredClaims
|
||||
}
|
||||
|
||||
// TokenIssuer signiert/verifiziert JWTs mit einem HMAC-Secret. Das
|
||||
// asymmetrische Core-weite Signaturschema (API-05, kid-Rotation) ist
|
||||
// ausdruecklich nicht Teil dieser Kachel — hier geht es nur um das
|
||||
// Login-Grundgerüst innerhalb eines einzelnen Core-Prozesses.
|
||||
type TokenIssuer struct {
|
||||
secret []byte
|
||||
}
|
||||
|
||||
func NewTokenIssuer(secret string) *TokenIssuer {
|
||||
return &TokenIssuer{secret: []byte(secret)}
|
||||
}
|
||||
|
||||
func (i *TokenIssuer) Issue(userID, tenantSlug string) (string, error) {
|
||||
now := time.Now()
|
||||
claims := Claims{
|
||||
UserID: userID,
|
||||
TenantSlug: tenantSlug,
|
||||
RegisteredClaims: jwt.RegisteredClaims{
|
||||
IssuedAt: jwt.NewNumericDate(now),
|
||||
ExpiresAt: jwt.NewNumericDate(now.Add(AccessTokenTTL)),
|
||||
},
|
||||
}
|
||||
token := jwt.NewWithClaims(jwt.SigningMethodHS256, claims)
|
||||
return token.SignedString(i.secret)
|
||||
}
|
||||
|
||||
// Verify prueft Signatur UND Ablauf (jwt.ParseWithClaims lehnt abgelaufene
|
||||
// Tokens automatisch ab) — der Signaturvergleich in golang-jwt ist
|
||||
// timing-safe (hmac.Equal).
|
||||
func (i *TokenIssuer) Verify(tokenString string) (*Claims, error) {
|
||||
claims := &Claims{}
|
||||
token, err := jwt.ParseWithClaims(tokenString, claims, func(t *jwt.Token) (interface{}, error) {
|
||||
if _, ok := t.Method.(*jwt.SigningMethodHMAC); !ok {
|
||||
return nil, ErrInvalidToken
|
||||
}
|
||||
return i.secret, nil
|
||||
})
|
||||
if err != nil || !token.Valid {
|
||||
return nil, ErrInvalidToken
|
||||
}
|
||||
return claims, nil
|
||||
}
|
||||
@@ -0,0 +1,82 @@
|
||||
package auth
|
||||
|
||||
import (
|
||||
"strings"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/golang-jwt/jwt/v5"
|
||||
)
|
||||
|
||||
func TestTokenIssueAndVerify(t *testing.T) {
|
||||
issuer := NewTokenIssuer("test-secret-nur-fuer-tests")
|
||||
|
||||
token, err := issuer.Issue("user-1", "acme")
|
||||
if err != nil {
|
||||
t.Fatalf("issue: %v", err)
|
||||
}
|
||||
|
||||
claims, err := issuer.Verify(token)
|
||||
if err != nil {
|
||||
t.Fatalf("verify: %v", err)
|
||||
}
|
||||
if claims.UserID != "user-1" || claims.TenantSlug != "acme" {
|
||||
t.Fatalf("claims unerwartet: %+v", claims)
|
||||
}
|
||||
}
|
||||
|
||||
// Pruefung 2: Token-Manipulationstest.
|
||||
func TestTokenVerify_RejectsManipulatedPayload(t *testing.T) {
|
||||
issuer := NewTokenIssuer("test-secret-nur-fuer-tests")
|
||||
token, err := issuer.Issue("user-1", "acme")
|
||||
if err != nil {
|
||||
t.Fatalf("issue: %v", err)
|
||||
}
|
||||
|
||||
parts := strings.Split(token, ".")
|
||||
if len(parts) != 3 {
|
||||
t.Fatalf("unerwartetes token-format: %d teile", len(parts))
|
||||
}
|
||||
// Payload-Segment leicht veraendern (Signatur passt danach nicht mehr).
|
||||
tampered := parts[0] + "." + parts[1] + "x" + "." + parts[2]
|
||||
|
||||
if _, err := issuer.Verify(tampered); err == nil {
|
||||
t.Fatal("erwartet fehler bei manipuliertem token, habe nil")
|
||||
}
|
||||
}
|
||||
|
||||
func TestTokenVerify_RejectsWrongSecret(t *testing.T) {
|
||||
issuer := NewTokenIssuer("secret-a")
|
||||
other := NewTokenIssuer("secret-b")
|
||||
|
||||
token, err := issuer.Issue("user-1", "acme")
|
||||
if err != nil {
|
||||
t.Fatalf("issue: %v", err)
|
||||
}
|
||||
if _, err := other.Verify(token); err == nil {
|
||||
t.Fatal("erwartet fehler bei falschem secret, habe nil")
|
||||
}
|
||||
}
|
||||
|
||||
// Pruefung 3: abgelaufenes Token erzwingt Neuanmeldung.
|
||||
func TestTokenVerify_RejectsExpiredToken(t *testing.T) {
|
||||
issuer := NewTokenIssuer("test-secret-nur-fuer-tests")
|
||||
|
||||
claims := Claims{
|
||||
UserID: "user-1",
|
||||
TenantSlug: "acme",
|
||||
RegisteredClaims: jwt.RegisteredClaims{
|
||||
IssuedAt: jwt.NewNumericDate(time.Now().Add(-2 * AccessTokenTTL)),
|
||||
ExpiresAt: jwt.NewNumericDate(time.Now().Add(-time.Minute)),
|
||||
},
|
||||
}
|
||||
tok := jwt.NewWithClaims(jwt.SigningMethodHS256, claims)
|
||||
expired, err := tok.SignedString([]byte("test-secret-nur-fuer-tests"))
|
||||
if err != nil {
|
||||
t.Fatalf("signieren: %v", err)
|
||||
}
|
||||
|
||||
if _, err := issuer.Verify(expired); err == nil {
|
||||
t.Fatal("erwartet fehler bei abgelaufenem token, habe nil")
|
||||
}
|
||||
}
|
||||
@@ -1,88 +0,0 @@
|
||||
package cfgservice
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"sync"
|
||||
"time"
|
||||
)
|
||||
|
||||
// DefaultCacheTTL ist die dokumentierte Cache-Invalidierungszeit
|
||||
// (Akzeptanzkriterium 2 / Pruefung 2 in diesem Ticket bezieht sich auf die
|
||||
// Aenderungsnachvollziehbarkeit — die Cache-Frist selbst folgt demselben
|
||||
// Muster wie internal/flag.DefaultCacheTTL).
|
||||
const DefaultCacheTTL = 5 * time.Second
|
||||
|
||||
type cacheEntry struct {
|
||||
value Value
|
||||
expiresAt time.Time
|
||||
}
|
||||
|
||||
// Service ist die Leseseite mit Vorrangregel (Akzeptanzkriterium 1:
|
||||
// Tenant-Override vor Global-Default) und lokalem TTL-Cache.
|
||||
type Service struct {
|
||||
store *Store
|
||||
ttl time.Duration
|
||||
|
||||
mu sync.RWMutex
|
||||
cache map[string]cacheEntry // Schluessel: key + "\x00" + tenantSlug
|
||||
}
|
||||
|
||||
func NewService(store *Store, ttl time.Duration) *Service {
|
||||
if ttl <= 0 {
|
||||
ttl = DefaultCacheTTL
|
||||
}
|
||||
return &Service{store: store, ttl: ttl, cache: make(map[string]cacheEntry)}
|
||||
}
|
||||
|
||||
func cacheKey(key, tenantSlug string) string {
|
||||
return key + "\x00" + tenantSlug
|
||||
}
|
||||
|
||||
// Resolve liefert den Konfigurationswert fuer einen Tenant: ein
|
||||
// Tenant-spezifischer Override hat Vorrang vor dem globalen Default
|
||||
// (Akzeptanzkriterium 1 / Pruefung 1). tenantSlug == "" wertet nur den
|
||||
// globalen Wert aus.
|
||||
func (s *Service) Resolve(ctx context.Context, tenantSlug, key string) (Value, error) {
|
||||
ck := cacheKey(key, tenantSlug)
|
||||
|
||||
s.mu.RLock()
|
||||
entry, exists := s.cache[ck]
|
||||
fresh := exists && time.Now().Before(entry.expiresAt)
|
||||
s.mu.RUnlock()
|
||||
if fresh {
|
||||
return entry.value, nil
|
||||
}
|
||||
|
||||
v, err := s.resolveUncached(ctx, tenantSlug, key)
|
||||
if err != nil {
|
||||
return Value{}, err
|
||||
}
|
||||
|
||||
s.mu.Lock()
|
||||
s.cache[ck] = cacheEntry{value: v, expiresAt: time.Now().Add(s.ttl)}
|
||||
s.mu.Unlock()
|
||||
return v, nil
|
||||
}
|
||||
|
||||
func (s *Service) resolveUncached(ctx context.Context, tenantSlug, key string) (Value, error) {
|
||||
if tenantSlug != "" {
|
||||
v, err := s.store.Get(ctx, key, tenantSlug)
|
||||
if err == nil {
|
||||
return v, nil
|
||||
}
|
||||
if !errors.Is(err, ErrNotFound) {
|
||||
return Value{}, err
|
||||
}
|
||||
}
|
||||
return s.store.Get(ctx, key, GlobalScope)
|
||||
}
|
||||
|
||||
// Invalidate erzwingt beim naechsten Resolve-Aufruf ein sofortiges Neuladen
|
||||
// fuer einen bestimmten (key, tenantSlug) statt auf den TTL-Ablauf zu warten
|
||||
// — analog internal/flag.Service.Invalidate.
|
||||
func (s *Service) Invalidate(key, tenantSlug string) {
|
||||
s.mu.Lock()
|
||||
delete(s.cache, cacheKey(key, tenantSlug))
|
||||
s.mu.Unlock()
|
||||
}
|
||||
@@ -1,114 +0,0 @@
|
||||
package cfgservice
|
||||
|
||||
import (
|
||||
"context"
|
||||
"testing"
|
||||
"time"
|
||||
)
|
||||
|
||||
// Akzeptanzkriterium 1 + Pruefung 1: Tenant-Override hat Vorrang vor
|
||||
// Global-Default, automatisiert getestet.
|
||||
func TestService_TenantOverrideTakesPrecedenceOverGlobal(t *testing.T) {
|
||||
store, cleanup := setupStoreTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
if _, err := store.Set(ctx, "test_precedence_key", GlobalScope, "global-wert"); err != nil {
|
||||
t.Fatalf("set global: %v", err)
|
||||
}
|
||||
if _, err := store.Set(ctx, "test_precedence_key", "test_acme", "tenant-wert"); err != nil {
|
||||
t.Fatalf("set tenant: %v", err)
|
||||
}
|
||||
|
||||
svc := NewService(store, time.Hour)
|
||||
|
||||
got, err := svc.Resolve(ctx, "test_acme", "test_precedence_key")
|
||||
if err != nil {
|
||||
t.Fatalf("resolve mit override: %v", err)
|
||||
}
|
||||
if got.Value != "tenant-wert" {
|
||||
t.Fatalf("erwartet tenant-override, habe %q", got.Value)
|
||||
}
|
||||
|
||||
gotOther, err := svc.Resolve(ctx, "test_anderer_tenant", "test_precedence_key")
|
||||
if err != nil {
|
||||
t.Fatalf("resolve ohne override: %v", err)
|
||||
}
|
||||
if gotOther.Value != "global-wert" {
|
||||
t.Fatalf("erwartet global-default fuer tenant ohne override, habe %q", gotOther.Value)
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 + Pruefung 2: Cache-Invalidierung nach
|
||||
// Konfigurationsaenderung innerhalb dokumentierter Zeit gemessen.
|
||||
func TestService_CacheInvalidationTiming(t *testing.T) {
|
||||
store, cleanup := setupStoreTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
const ttl = 150 * time.Millisecond
|
||||
if _, err := store.Set(ctx, "test_ttl_key", GlobalScope, "alt"); err != nil {
|
||||
t.Fatalf("set: %v", err)
|
||||
}
|
||||
svc := NewService(store, ttl)
|
||||
|
||||
v, err := svc.Resolve(ctx, "", "test_ttl_key")
|
||||
if err != nil {
|
||||
t.Fatalf("resolve: %v", err)
|
||||
}
|
||||
if v.Value != "alt" {
|
||||
t.Fatalf("erwartet 'alt', habe %q", v.Value)
|
||||
}
|
||||
|
||||
changedAt := time.Now()
|
||||
if _, err := store.Set(ctx, "test_ttl_key", GlobalScope, "neu"); err != nil {
|
||||
t.Fatalf("set: %v", err)
|
||||
}
|
||||
|
||||
v, err = svc.Resolve(ctx, "", "test_ttl_key")
|
||||
if err != nil {
|
||||
t.Fatalf("resolve direkt nach aenderung: %v", err)
|
||||
}
|
||||
if v.Value != "alt" {
|
||||
t.Fatalf("cache haette den alten wert liefern sollen, habe %q", v.Value)
|
||||
}
|
||||
|
||||
deadline := changedAt.Add(ttl + 100*time.Millisecond)
|
||||
for time.Now().Before(deadline) {
|
||||
v, err := svc.Resolve(ctx, "", "test_ttl_key")
|
||||
if err != nil {
|
||||
t.Fatalf("resolve: %v", err)
|
||||
}
|
||||
if v.Value == "neu" {
|
||||
t.Logf("aenderung wurde nach %s wirksam (ziel: innerhalb %s + toleranz)", time.Since(changedAt), ttl)
|
||||
return
|
||||
}
|
||||
time.Sleep(10 * time.Millisecond)
|
||||
}
|
||||
t.Fatalf("aenderung wurde nicht innerhalb von %s wirksam", deadline.Sub(changedAt))
|
||||
}
|
||||
|
||||
func TestService_InvalidateForcesImmediateRefresh(t *testing.T) {
|
||||
store, cleanup := setupStoreTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
if _, err := store.Set(ctx, "test_invalidate_key", GlobalScope, "alt"); err != nil {
|
||||
t.Fatalf("set: %v", err)
|
||||
}
|
||||
svc := NewService(store, time.Hour)
|
||||
_, _ = svc.Resolve(ctx, "", "test_invalidate_key")
|
||||
|
||||
if _, err := store.Set(ctx, "test_invalidate_key", GlobalScope, "neu"); err != nil {
|
||||
t.Fatalf("set: %v", err)
|
||||
}
|
||||
svc.Invalidate("test_invalidate_key", "")
|
||||
|
||||
v, err := svc.Resolve(ctx, "", "test_invalidate_key")
|
||||
if err != nil {
|
||||
t.Fatalf("resolve: %v", err)
|
||||
}
|
||||
if v.Value != "neu" {
|
||||
t.Fatalf("erwartet sofort sichtbaren neuen wert nach Invalidate, habe %q", v.Value)
|
||||
}
|
||||
}
|
||||
@@ -1,129 +0,0 @@
|
||||
// Package cfgservice implementiert Core CFG-01: den zentralen Dienst fuer
|
||||
// globale und tenant-spezifische Konfigurationswerte mit Versionierung und
|
||||
// Cache-Invalidierung. Andere Module lesen Konfiguration AUSSCHLIESSLICH
|
||||
// ueber dieses Paket (Akzeptanzkriterium 3), niemals ueber eigene Tabellen.
|
||||
package cfgservice
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"fmt"
|
||||
|
||||
"github.com/jackc/pgx/v5"
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
// GlobalScope ist der reservierte Scope-Wert fuer globale Defaults — jeder
|
||||
// andere Scope-Wert ist ein Tenant-Slug (Akzeptanzkriterium 1).
|
||||
const GlobalScope = "global"
|
||||
|
||||
var ErrNotFound = errors.New("cfgservice: kein wert fuer diesen key gefunden")
|
||||
|
||||
type Value struct {
|
||||
Key string
|
||||
Scope string
|
||||
Value string
|
||||
Version int
|
||||
}
|
||||
|
||||
type HistoryEntry struct {
|
||||
Key string
|
||||
Scope string
|
||||
Value string
|
||||
Version int
|
||||
}
|
||||
|
||||
// Store ist die Schreib-/Verwaltungsseite. Set schreibt IMMER sowohl den
|
||||
// aktuellen Stand (config_values) als auch einen Historieneintrag
|
||||
// (config_value_history) in derselben Transaktion — eine Aenderung ohne
|
||||
// Versionshistorie ist strukturell ausgeschlossen (Akzeptanzkriterium 2).
|
||||
type Store struct {
|
||||
pool *pgxpool.Pool
|
||||
}
|
||||
|
||||
func NewStore(pool *pgxpool.Pool) *Store {
|
||||
return &Store{pool: pool}
|
||||
}
|
||||
|
||||
// Set schreibt einen neuen Wert fuer (key, scope) und erhoeht die Version um 1
|
||||
// (Version 1 bei erstmaligem Setzen).
|
||||
func (s *Store) Set(ctx context.Context, key, scope, value string) (Value, error) {
|
||||
if scope == "" {
|
||||
return Value{}, errors.New("cfgservice: scope darf nicht leer sein")
|
||||
}
|
||||
|
||||
tx, err := s.pool.Begin(ctx)
|
||||
if err != nil {
|
||||
return Value{}, fmt.Errorf("transaktion starten: %w", err)
|
||||
}
|
||||
defer func() { _ = tx.Rollback(ctx) }()
|
||||
|
||||
var currentVersion int
|
||||
err = tx.QueryRow(ctx, `SELECT version FROM config_values WHERE key = $1 AND scope = $2`, key, scope).Scan(¤tVersion)
|
||||
if err != nil && !errors.Is(err, pgx.ErrNoRows) {
|
||||
return Value{}, fmt.Errorf("aktuelle version lesen: %w", err)
|
||||
}
|
||||
newVersion := currentVersion + 1
|
||||
|
||||
if _, err := tx.Exec(ctx, `
|
||||
INSERT INTO config_values (key, scope, value, version, updated_at)
|
||||
VALUES ($1, $2, $3, $4, now())
|
||||
ON CONFLICT (key, scope) DO UPDATE SET value = $3, version = $4, updated_at = now()
|
||||
`, key, scope, value, newVersion); err != nil {
|
||||
return Value{}, fmt.Errorf("wert speichern: %w", err)
|
||||
}
|
||||
|
||||
if _, err := tx.Exec(ctx, `
|
||||
INSERT INTO config_value_history (key, scope, value, version, changed_at)
|
||||
VALUES ($1, $2, $3, $4, now())
|
||||
`, key, scope, value, newVersion); err != nil {
|
||||
return Value{}, fmt.Errorf("historie schreiben: %w", err)
|
||||
}
|
||||
|
||||
if err := tx.Commit(ctx); err != nil {
|
||||
return Value{}, fmt.Errorf("transaktion committen: %w", err)
|
||||
}
|
||||
|
||||
return Value{Key: key, Scope: scope, Value: value, Version: newVersion}, nil
|
||||
}
|
||||
|
||||
// Get liefert den Wert fuer GENAU EINEN Scope (kein Vorrang-Fallback) — die
|
||||
// Vorrangregel (Tenant vor Global) lebt bewusst in Service.Resolve, damit
|
||||
// Store rein CRUD bleibt.
|
||||
func (s *Store) Get(ctx context.Context, key, scope string) (Value, error) {
|
||||
var v Value
|
||||
v.Key, v.Scope = key, scope
|
||||
err := s.pool.QueryRow(ctx, `
|
||||
SELECT value, version FROM config_values WHERE key = $1 AND scope = $2
|
||||
`, key, scope).Scan(&v.Value, &v.Version)
|
||||
if err != nil {
|
||||
if errors.Is(err, pgx.ErrNoRows) {
|
||||
return Value{}, ErrNotFound
|
||||
}
|
||||
return Value{}, fmt.Errorf("wert lesen: %w", err)
|
||||
}
|
||||
return v, nil
|
||||
}
|
||||
|
||||
// History liefert die vollstaendige Versionshistorie eines (key, scope) in
|
||||
// aufsteigender Reihenfolge (Akzeptanzkriterium 2 / Pruefung 3).
|
||||
func (s *Store) History(ctx context.Context, key, scope string) ([]HistoryEntry, error) {
|
||||
rows, err := s.pool.Query(ctx, `
|
||||
SELECT key, scope, value, version FROM config_value_history
|
||||
WHERE key = $1 AND scope = $2 ORDER BY version
|
||||
`, key, scope)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("historie abfragen: %w", err)
|
||||
}
|
||||
defer rows.Close()
|
||||
|
||||
var out []HistoryEntry
|
||||
for rows.Next() {
|
||||
var h HistoryEntry
|
||||
if err := rows.Scan(&h.Key, &h.Scope, &h.Value, &h.Version); err != nil {
|
||||
return nil, fmt.Errorf("historieneintrag lesen: %w", err)
|
||||
}
|
||||
out = append(out, h)
|
||||
}
|
||||
return out, rows.Err()
|
||||
}
|
||||
@@ -1,119 +0,0 @@
|
||||
package cfgservice
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"os"
|
||||
"testing"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
func setupStoreTest(t *testing.T) (*Store, func()) {
|
||||
t.Helper()
|
||||
adminDSN := os.Getenv("TEST_ADMIN_DSN")
|
||||
if adminDSN == "" {
|
||||
t.Skip("TEST_ADMIN_DSN nicht gesetzt, Integrationstest uebersprungen")
|
||||
}
|
||||
ctx := context.Background()
|
||||
|
||||
pool, err := pgxpool.New(ctx, adminDSN)
|
||||
if err != nil {
|
||||
t.Fatalf("pool: %v", err)
|
||||
}
|
||||
if _, err := pool.Exec(ctx, `
|
||||
CREATE TABLE IF NOT EXISTS config_values (
|
||||
key TEXT NOT NULL,
|
||||
scope TEXT NOT NULL CHECK (scope <> ''),
|
||||
value TEXT NOT NULL,
|
||||
version INT NOT NULL,
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
PRIMARY KEY (key, scope)
|
||||
);
|
||||
CREATE TABLE IF NOT EXISTS config_value_history (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
key TEXT NOT NULL,
|
||||
scope TEXT NOT NULL,
|
||||
value TEXT NOT NULL,
|
||||
version INT NOT NULL,
|
||||
changed_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
`); err != nil {
|
||||
t.Fatalf("schema: %v", err)
|
||||
}
|
||||
|
||||
cleanup := func() {
|
||||
_, _ = pool.Exec(ctx, `DELETE FROM config_value_history WHERE key LIKE 'test\_%' ESCAPE '\'`)
|
||||
_, _ = pool.Exec(ctx, `DELETE FROM config_values WHERE key LIKE 'test\_%' ESCAPE '\'`)
|
||||
pool.Close()
|
||||
}
|
||||
return NewStore(pool), cleanup
|
||||
}
|
||||
|
||||
func TestStore_SetIncrementsVersion(t *testing.T) {
|
||||
store, cleanup := setupStoreTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
v1, err := store.Set(ctx, "test_key", GlobalScope, "erster-wert")
|
||||
if err != nil {
|
||||
t.Fatalf("set 1: %v", err)
|
||||
}
|
||||
if v1.Version != 1 {
|
||||
t.Fatalf("erwartet version 1, habe %d", v1.Version)
|
||||
}
|
||||
|
||||
v2, err := store.Set(ctx, "test_key", GlobalScope, "zweiter-wert")
|
||||
if err != nil {
|
||||
t.Fatalf("set 2: %v", err)
|
||||
}
|
||||
if v2.Version != 2 {
|
||||
t.Fatalf("erwartet version 2, habe %d", v2.Version)
|
||||
}
|
||||
|
||||
got, err := store.Get(ctx, "test_key", GlobalScope)
|
||||
if err != nil {
|
||||
t.Fatalf("get: %v", err)
|
||||
}
|
||||
if got.Value != "zweiter-wert" || got.Version != 2 {
|
||||
t.Fatalf("aktueller wert unerwartet: %+v", got)
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 + Pruefung 3: Versionierungshistorie ueber mehrere
|
||||
// Aenderungen hinweg nachvollzogen.
|
||||
func TestStore_HistoryTracksAllChanges(t *testing.T) {
|
||||
store, cleanup := setupStoreTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
values := []string{"v1", "v2", "v3"}
|
||||
for _, v := range values {
|
||||
if _, err := store.Set(ctx, "test_history_key", GlobalScope, v); err != nil {
|
||||
t.Fatalf("set %q: %v", v, err)
|
||||
}
|
||||
}
|
||||
|
||||
history, err := store.History(ctx, "test_history_key", GlobalScope)
|
||||
if err != nil {
|
||||
t.Fatalf("history: %v", err)
|
||||
}
|
||||
if len(history) != 3 {
|
||||
t.Fatalf("erwartet 3 historieneintraege, habe %d", len(history))
|
||||
}
|
||||
for i, h := range history {
|
||||
if h.Version != i+1 || h.Value != values[i] {
|
||||
t.Fatalf("historieneintrag[%d] unerwartet: %+v", i, h)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
func TestStore_GetUnknownKeyReturnsNotFound(t *testing.T) {
|
||||
store, cleanup := setupStoreTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
if _, err := store.Get(ctx, "test_nie_gesetzt", GlobalScope); !errors.Is(err, ErrNotFound) {
|
||||
t.Fatalf("erwartet ErrNotFound, habe %v", err)
|
||||
}
|
||||
}
|
||||
@@ -1,87 +0,0 @@
|
||||
// Package flag implementiert Core LIC-02: einen Feature-Flag-Dienst mit
|
||||
// Strategien (global an/aus, Prozentsatz, Tenant-Zielgruppe) als Kernfunktion
|
||||
// des Core-Dienstes selbst — keine zusaetzliche Infrastruktur (Unleash-Server
|
||||
// + eigene DB), siehe "bewusst vermeiden" im LIC-02-Ticket.
|
||||
package flag
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"fmt"
|
||||
"hash/fnv"
|
||||
|
||||
"github.com/jackc/pgx/v5"
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
var ErrNotFound = errors.New("flag: nicht gefunden")
|
||||
|
||||
// Flag ist die zentrale Definition — Auswertung (Evaluate) ist bewusst davon
|
||||
// getrennt (Unleash-Prinzip: Flag-Verwaltung vs. Flag-Auswertung).
|
||||
type Flag struct {
|
||||
Key string
|
||||
Enabled bool
|
||||
RolloutPercentage int
|
||||
TargetTenantSlugs []string
|
||||
}
|
||||
|
||||
// Store ist die Verwaltungsseite (Admin): Flags definieren/lesen.
|
||||
type Store struct {
|
||||
pool *pgxpool.Pool
|
||||
}
|
||||
|
||||
func NewStore(pool *pgxpool.Pool) *Store {
|
||||
return &Store{pool: pool}
|
||||
}
|
||||
|
||||
func (s *Store) Set(ctx context.Context, f Flag) error {
|
||||
if f.TargetTenantSlugs == nil {
|
||||
f.TargetTenantSlugs = []string{} // pgx uebertraegt ein nil-Slice sonst als SQL NULL statt leerem Array.
|
||||
}
|
||||
_, err := s.pool.Exec(ctx, `
|
||||
INSERT INTO feature_flags (key, enabled, rollout_percentage, target_tenant_slugs, updated_at)
|
||||
VALUES ($1, $2, $3, $4, now())
|
||||
ON CONFLICT (key) DO UPDATE SET
|
||||
enabled = $2, rollout_percentage = $3, target_tenant_slugs = $4, updated_at = now()
|
||||
`, f.Key, f.Enabled, f.RolloutPercentage, f.TargetTenantSlugs)
|
||||
if err != nil {
|
||||
return fmt.Errorf("flag speichern: %w", err)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
func (s *Store) Get(ctx context.Context, key string) (Flag, error) {
|
||||
var f Flag
|
||||
row := s.pool.QueryRow(ctx, `
|
||||
SELECT key, enabled, rollout_percentage, target_tenant_slugs
|
||||
FROM feature_flags WHERE key = $1
|
||||
`, key)
|
||||
if err := row.Scan(&f.Key, &f.Enabled, &f.RolloutPercentage, &f.TargetTenantSlugs); err != nil {
|
||||
if errors.Is(err, pgx.ErrNoRows) {
|
||||
return Flag{}, ErrNotFound
|
||||
}
|
||||
return Flag{}, fmt.Errorf("flag lesen: %w", err)
|
||||
}
|
||||
return f, nil
|
||||
}
|
||||
|
||||
// evaluate wendet die Strategien in fester Reihenfolge an: globaler
|
||||
// An/Aus-Schalter zuerst, dann Tenant-Zielgruppe, dann Prozentsatz-Rollout.
|
||||
// Ein unbekannter/nicht getroffener Fall ergibt false — Fail-Safe-Default,
|
||||
// kein Feature wird versehentlich aktiv.
|
||||
func evaluate(f Flag, tenantSlug string) bool {
|
||||
if f.Enabled {
|
||||
return true
|
||||
}
|
||||
for _, target := range f.TargetTenantSlugs {
|
||||
if target == tenantSlug {
|
||||
return true
|
||||
}
|
||||
}
|
||||
if f.RolloutPercentage > 0 {
|
||||
h := fnv.New32a()
|
||||
_, _ = h.Write([]byte(f.Key + "|" + tenantSlug))
|
||||
return int(h.Sum32()%100) < f.RolloutPercentage
|
||||
}
|
||||
return false
|
||||
}
|
||||
@@ -1,43 +0,0 @@
|
||||
package flag
|
||||
|
||||
import "testing"
|
||||
|
||||
func TestEvaluate_GlobalEnabled(t *testing.T) {
|
||||
f := Flag{Key: "k", Enabled: true}
|
||||
if !evaluate(f, "irgendein-tenant") {
|
||||
t.Fatal("global aktiviertes flag sollte fuer jeden tenant true liefern")
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 1 + Pruefung 2: Zielgruppen-Strategie.
|
||||
func TestEvaluate_TargetTenantStrategy(t *testing.T) {
|
||||
f := Flag{Key: "k", Enabled: false, TargetTenantSlugs: []string{"acme"}}
|
||||
if !evaluate(f, "acme") {
|
||||
t.Fatal("erwartet true fuer tenant in zielgruppe")
|
||||
}
|
||||
if evaluate(f, "globex") {
|
||||
t.Fatal("erwartet false fuer tenant ausserhalb der zielgruppe")
|
||||
}
|
||||
}
|
||||
|
||||
func TestEvaluate_RolloutPercentageBoundaries(t *testing.T) {
|
||||
full := Flag{Key: "k", RolloutPercentage: 100}
|
||||
if !evaluate(full, "beliebiger-tenant-1") || !evaluate(full, "beliebiger-tenant-2") {
|
||||
t.Fatal("100% rollout sollte immer true liefern")
|
||||
}
|
||||
|
||||
none := Flag{Key: "k", RolloutPercentage: 0}
|
||||
if evaluate(none, "beliebiger-tenant") {
|
||||
t.Fatal("0% rollout ohne enabled/zielgruppe sollte false liefern")
|
||||
}
|
||||
}
|
||||
|
||||
func TestEvaluate_RolloutIsDeterministicPerTenant(t *testing.T) {
|
||||
f := Flag{Key: "k", RolloutPercentage: 50}
|
||||
first := evaluate(f, "stabiler-tenant")
|
||||
for i := 0; i < 5; i++ {
|
||||
if evaluate(f, "stabiler-tenant") != first {
|
||||
t.Fatal("rollout-auswertung sollte fuer denselben tenant/key stabil sein")
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -1,87 +0,0 @@
|
||||
package flag
|
||||
|
||||
import (
|
||||
"context"
|
||||
"log/slog"
|
||||
"sync"
|
||||
"time"
|
||||
)
|
||||
|
||||
// DefaultCacheTTL ist die dokumentierte Cache-Invalidierungszeit
|
||||
// (Akzeptanzkriterium 2/3): eine Aenderung wirkt spaetestens nach dieser
|
||||
// Zeit auf allen Core-Instanzen, ohne dass ein Dienst neu gestartet werden
|
||||
// muss (Akzeptanzkriterium 3).
|
||||
const DefaultCacheTTL = 5 * time.Second
|
||||
|
||||
type cacheEntry struct {
|
||||
flag Flag
|
||||
expiresAt time.Time
|
||||
}
|
||||
|
||||
// Service ist die Auswertungsseite (SDK/Client-Analogon zu Unleash) mit
|
||||
// lokalem TTL-Cache. Bewusst getrennt von Store (Verwaltung).
|
||||
type Service struct {
|
||||
store *Store
|
||||
ttl time.Duration
|
||||
|
||||
mu sync.RWMutex
|
||||
cache map[string]cacheEntry
|
||||
}
|
||||
|
||||
func NewService(store *Store, ttl time.Duration) *Service {
|
||||
if ttl <= 0 {
|
||||
ttl = DefaultCacheTTL
|
||||
}
|
||||
return &Service{store: store, ttl: ttl, cache: make(map[string]cacheEntry)}
|
||||
}
|
||||
|
||||
// IsEnabled wertet ein Flag fuer einen Tenant aus. Liefert IMMER einen
|
||||
// bool ohne Fehlerwert — ein nicht erreichbarer Flag-Dienst darf abhaengige
|
||||
// Aufrufer nicht zum Absturz bringen oder zu Fehlerbehandlungscode zwingen,
|
||||
// der leicht vergessen wird (Akzeptanzkriterium 3 / Pruefung 3: dokumentiertes
|
||||
// Fallback-Verhalten = false, ggf. aus dem zuletzt bekannten Zwischenspeicher).
|
||||
func (s *Service) IsEnabled(ctx context.Context, tenantSlug, key string) bool {
|
||||
f, ok := s.resolve(ctx, key)
|
||||
if !ok {
|
||||
return false
|
||||
}
|
||||
return evaluate(f, tenantSlug)
|
||||
}
|
||||
|
||||
func (s *Service) resolve(ctx context.Context, key string) (Flag, bool) {
|
||||
s.mu.RLock()
|
||||
entry, exists := s.cache[key]
|
||||
fresh := exists && time.Now().Before(entry.expiresAt)
|
||||
s.mu.RUnlock()
|
||||
if fresh {
|
||||
return entry.flag, true
|
||||
}
|
||||
|
||||
f, err := s.store.Get(ctx, key)
|
||||
if err != nil {
|
||||
if exists {
|
||||
slog.Warn("feature-flag-dienst nicht erreichbar, nutze zwischengespeicherten stand",
|
||||
"flag_key", key, "error", err)
|
||||
return entry.flag, true
|
||||
}
|
||||
slog.Warn("feature-flag-dienst nicht erreichbar, kein zwischengespeicherter stand vorhanden, fallback: deaktiviert",
|
||||
"flag_key", key, "error", err)
|
||||
return Flag{}, false
|
||||
}
|
||||
|
||||
s.mu.Lock()
|
||||
s.cache[key] = cacheEntry{flag: f, expiresAt: time.Now().Add(s.ttl)}
|
||||
s.mu.Unlock()
|
||||
return f, true
|
||||
}
|
||||
|
||||
// Invalidate erzwingt beim naechsten IsEnabled-Aufruf ein sofortiges Neuladen
|
||||
// aus der Datenbank statt auf den TTL-Ablauf zu warten — wird nach Store.Set
|
||||
// auf derselben Instanz aufgerufen, damit der Schreiber die eigene Aenderung
|
||||
// ohne Wartezeit sieht. Andere Core-Instanzen sehen sie spaetestens nach
|
||||
// DefaultCacheTTL (siehe Akzeptanzkriterium 3).
|
||||
func (s *Service) Invalidate(key string) {
|
||||
s.mu.Lock()
|
||||
delete(s.cache, key)
|
||||
s.mu.Unlock()
|
||||
}
|
||||
@@ -1,179 +0,0 @@
|
||||
package flag
|
||||
|
||||
import (
|
||||
"context"
|
||||
"os"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
func setupFlagStoreTest(t *testing.T) (*Store, func()) {
|
||||
t.Helper()
|
||||
adminDSN := os.Getenv("TEST_ADMIN_DSN")
|
||||
if adminDSN == "" {
|
||||
t.Skip("TEST_ADMIN_DSN nicht gesetzt, Integrationstest uebersprungen")
|
||||
}
|
||||
ctx := context.Background()
|
||||
|
||||
pool, err := pgxpool.New(ctx, adminDSN)
|
||||
if err != nil {
|
||||
t.Fatalf("pool: %v", err)
|
||||
}
|
||||
if _, err := pool.Exec(ctx, `
|
||||
CREATE TABLE IF NOT EXISTS feature_flags (
|
||||
key TEXT PRIMARY KEY,
|
||||
enabled BOOLEAN NOT NULL DEFAULT false,
|
||||
rollout_percentage INT NOT NULL DEFAULT 0,
|
||||
target_tenant_slugs TEXT[] NOT NULL DEFAULT '{}',
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
)`); err != nil {
|
||||
t.Fatalf("schema: %v", err)
|
||||
}
|
||||
|
||||
cleanup := func() {
|
||||
_, _ = pool.Exec(ctx, `DELETE FROM feature_flags WHERE key LIKE 'test\_%' ESCAPE '\'`)
|
||||
pool.Close()
|
||||
}
|
||||
return NewStore(pool), cleanup
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 1 + Pruefung 2: Zielgruppen-Strategie liefert im Test
|
||||
// die erwartete Auswertung.
|
||||
func TestService_TargetTenantStrategy(t *testing.T) {
|
||||
store, cleanup := setupFlagStoreTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
if err := store.Set(ctx, Flag{Key: "test_target_flag", TargetTenantSlugs: []string{"acme"}}); err != nil {
|
||||
t.Fatalf("set: %v", err)
|
||||
}
|
||||
svc := NewService(store, time.Hour)
|
||||
|
||||
if !svc.IsEnabled(ctx, "acme", "test_target_flag") {
|
||||
t.Fatal("erwartet true fuer tenant in zielgruppe")
|
||||
}
|
||||
if svc.IsEnabled(ctx, "globex", "test_target_flag") {
|
||||
t.Fatal("erwartet false fuer tenant ausserhalb der zielgruppe")
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 + 3 + Pruefung 1: Flag-Aenderung wirkt innerhalb der
|
||||
// dokumentierten Cache-Invalidierungszeit, automatisiert gemessen.
|
||||
func TestService_CacheInvalidationTiming(t *testing.T) {
|
||||
store, cleanup := setupFlagStoreTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
const ttl = 150 * time.Millisecond
|
||||
if err := store.Set(ctx, Flag{Key: "test_ttl_flag", Enabled: false}); err != nil {
|
||||
t.Fatalf("set: %v", err)
|
||||
}
|
||||
svc := NewService(store, ttl)
|
||||
|
||||
if svc.IsEnabled(ctx, "acme", "test_ttl_flag") {
|
||||
t.Fatal("erwartet false vor der aenderung")
|
||||
}
|
||||
|
||||
// Aenderung "auf einer anderen instanz" simulieren: direkt ueber den
|
||||
// Store, ohne svc.Invalidate aufzurufen.
|
||||
changedAt := time.Now()
|
||||
if err := store.Set(ctx, Flag{Key: "test_ttl_flag", Enabled: true}); err != nil {
|
||||
t.Fatalf("set: %v", err)
|
||||
}
|
||||
|
||||
// Sofort danach sollte der Cache noch den alten Stand liefern.
|
||||
if svc.IsEnabled(ctx, "acme", "test_ttl_flag") {
|
||||
t.Fatal("cache haette den alten (false) stand liefern sollen, direkt nach der aenderung")
|
||||
}
|
||||
|
||||
deadline := changedAt.Add(ttl + 100*time.Millisecond)
|
||||
for time.Now().Before(deadline) {
|
||||
if svc.IsEnabled(ctx, "acme", "test_ttl_flag") {
|
||||
elapsed := time.Since(changedAt)
|
||||
t.Logf("aenderung wurde nach %s wirksam (ziel: innerhalb %s + toleranz)", elapsed, ttl)
|
||||
return
|
||||
}
|
||||
time.Sleep(10 * time.Millisecond)
|
||||
}
|
||||
t.Fatalf("aenderung wurde nicht innerhalb von %s wirksam", deadline.Sub(changedAt))
|
||||
}
|
||||
|
||||
func TestService_InvalidateForcesImmediateRefresh(t *testing.T) {
|
||||
store, cleanup := setupFlagStoreTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
if err := store.Set(ctx, Flag{Key: "test_invalidate_flag", Enabled: false}); err != nil {
|
||||
t.Fatalf("set: %v", err)
|
||||
}
|
||||
svc := NewService(store, time.Hour) // lange TTL, damit Invalidate den unterschied macht
|
||||
_ = svc.IsEnabled(ctx, "acme", "test_invalidate_flag")
|
||||
|
||||
if err := store.Set(ctx, Flag{Key: "test_invalidate_flag", Enabled: true}); err != nil {
|
||||
t.Fatalf("set: %v", err)
|
||||
}
|
||||
svc.Invalidate("test_invalidate_flag")
|
||||
|
||||
if !svc.IsEnabled(ctx, "acme", "test_invalidate_flag") {
|
||||
t.Fatal("erwartet sofort sichtbaren neuen stand nach Invalidate")
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 3 + Pruefung 3: Ausfall des Flag-Dienstes fuehrt zu
|
||||
// dokumentiertem Fallback-Verhalten, nicht zum Absturz.
|
||||
func TestService_FallsBackOnStoreFailure(t *testing.T) {
|
||||
store, cleanup := setupFlagStoreTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
if err := store.Set(ctx, Flag{Key: "test_fallback_flag", Enabled: true}); err != nil {
|
||||
t.Fatalf("set: %v", err)
|
||||
}
|
||||
svc := NewService(store, time.Hour)
|
||||
|
||||
// Cache vorwaermen, waehrend die DB noch erreichbar ist.
|
||||
if !svc.IsEnabled(ctx, "acme", "test_fallback_flag") {
|
||||
t.Fatal("erwartet true bei funktionierender db")
|
||||
}
|
||||
|
||||
brokenPool, err := pgxpool.New(ctx, "postgresql://nonexistent-host-fuer-test:5432/x?connect_timeout=1")
|
||||
if err != nil {
|
||||
t.Fatalf("broken pool erstellen (sollte nicht sofort verbinden): %v", err)
|
||||
}
|
||||
brokenStore := NewStore(brokenPool)
|
||||
|
||||
svcWithCache := NewService(brokenStore, time.Nanosecond) // TTL sofort abgelaufen, erzwingt reload-versuch
|
||||
svcWithCache.mu.Lock()
|
||||
svcWithCache.cache["test_fallback_flag"] = cacheEntry{
|
||||
flag: Flag{Key: "test_fallback_flag", Enabled: true},
|
||||
expiresAt: time.Now().Add(-time.Hour), // bereits abgelaufen
|
||||
}
|
||||
svcWithCache.mu.Unlock()
|
||||
|
||||
func() {
|
||||
defer func() {
|
||||
if r := recover(); r != nil {
|
||||
t.Fatalf("IsEnabled hat gepanict statt einen fallback zu liefern: %v", r)
|
||||
}
|
||||
}()
|
||||
if !svcWithCache.IsEnabled(ctx, "acme", "test_fallback_flag") {
|
||||
t.Fatal("erwartet fallback auf zwischengespeicherten (true) stand bei db-ausfall")
|
||||
}
|
||||
}()
|
||||
|
||||
// Voellig frischer Dienst ohne jeglichen cache + kaputte db -> sicherer
|
||||
// default false, kein absturz.
|
||||
freshSvc := NewService(brokenStore, time.Hour)
|
||||
func() {
|
||||
defer func() {
|
||||
if r := recover(); r != nil {
|
||||
t.Fatalf("IsEnabled hat gepanict: %v", r)
|
||||
}
|
||||
}()
|
||||
if freshSvc.IsEnabled(ctx, "acme", "test_fallback_flag") {
|
||||
t.Fatal("erwartet fail-safe false ohne cache und mit kaputter db")
|
||||
}
|
||||
}()
|
||||
}
|
||||
@@ -1,25 +0,0 @@
|
||||
package health
|
||||
|
||||
import (
|
||||
"context"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
// DatabaseChecker prueft die tatsaechliche Erreichbarkeit der Datenbank
|
||||
// (Ping) — nicht nur, ob der Pool existiert.
|
||||
func DatabaseChecker(pool *pgxpool.Pool) CheckerFunc {
|
||||
return func(ctx context.Context) error {
|
||||
return pool.Ping(ctx)
|
||||
}
|
||||
}
|
||||
|
||||
// QueueChecker prueft, dass die Postgres-basierte Job-Queue (siehe CFG-02)
|
||||
// tatsaechlich abfragbar ist — eine eigene, benannte Abhaengigkeit neben der
|
||||
// reinen DB-Erreichbarkeit (Akzeptanzkriterium 1).
|
||||
func QueueChecker(pool *pgxpool.Pool) CheckerFunc {
|
||||
return func(ctx context.Context) error {
|
||||
_, err := pool.Exec(ctx, `SELECT 1`)
|
||||
return err
|
||||
}
|
||||
}
|
||||
@@ -1,49 +0,0 @@
|
||||
package health
|
||||
|
||||
import (
|
||||
"encoding/json"
|
||||
"net/http"
|
||||
)
|
||||
|
||||
// LivenessHandler beantwortet IMMER "lebt", solange der Prozess ueberhaupt
|
||||
// HTTP-Anfragen verarbeiten kann — prueft bewusst KEINE externen
|
||||
// Abhaengigkeiten (Akzeptanzkriterium 2: Liveness und Readiness getrennt).
|
||||
// Ein Datenbankausfall darf die Liveness nicht auf "tot" setzen, sonst
|
||||
// wuerde eine Orchestrierung (z.B. systemd/Kubernetes) den Prozess grundlos
|
||||
// neu starten, obwohl nur eine Abhaengigkeit ausgefallen ist.
|
||||
func LivenessHandler(w http.ResponseWriter, r *http.Request) {
|
||||
writeStatus(w, http.StatusOK, map[string]any{"status": "alive"})
|
||||
}
|
||||
|
||||
// ReadinessHandler prueft ALLE registrierten Abhaengigkeiten
|
||||
// (Akzeptanzkriterium 1) und liefert 503, sobald eine davon fehlschlaegt
|
||||
// (Akzeptanzkriterium 3) — unterscheidet sich damit nachweislich von
|
||||
// LivenessHandler im Fehlerfall (Akzeptanzkriterium 2 / Pruefung 3).
|
||||
func (r *Registry) ReadinessHandler() http.HandlerFunc {
|
||||
return func(w http.ResponseWriter, req *http.Request) {
|
||||
ready, results := r.CheckAll(req.Context())
|
||||
|
||||
body := map[string]any{
|
||||
"status": statusText(ready),
|
||||
"checks": results,
|
||||
}
|
||||
status := http.StatusOK
|
||||
if !ready {
|
||||
status = http.StatusServiceUnavailable
|
||||
}
|
||||
writeStatus(w, status, body)
|
||||
}
|
||||
}
|
||||
|
||||
func statusText(ready bool) string {
|
||||
if ready {
|
||||
return "ready"
|
||||
}
|
||||
return "not_ready"
|
||||
}
|
||||
|
||||
func writeStatus(w http.ResponseWriter, status int, body map[string]any) {
|
||||
w.Header().Set("Content-Type", "application/json")
|
||||
w.WriteHeader(status)
|
||||
_ = json.NewEncoder(w).Encode(body)
|
||||
}
|
||||
@@ -1,86 +0,0 @@
|
||||
// Package health implementiert Core OPS-01: Health-/Readiness-Endpunkte, die
|
||||
// echte Abhaengigkeiten (DB, Job-Queue) statt nur den Prozessstatus pruefen
|
||||
// — wiederverwendbar von Core UND jedem registrierten Modul (siehe API-02),
|
||||
// nicht nur von Core selbst.
|
||||
package health
|
||||
|
||||
import (
|
||||
"context"
|
||||
"time"
|
||||
)
|
||||
|
||||
// Checker prueft EINE Abhaengigkeit (z.B. Datenbank, Job-Queue).
|
||||
type Checker interface {
|
||||
Check(ctx context.Context) error
|
||||
}
|
||||
|
||||
type CheckerFunc func(ctx context.Context) error
|
||||
|
||||
func (f CheckerFunc) Check(ctx context.Context) error { return f(ctx) }
|
||||
|
||||
// DefaultCheckTimeout begrenzt, wie lange EIN einzelner Check maximal
|
||||
// dauern darf, bevor er als fehlgeschlagen gilt — verhindert, dass ein
|
||||
// haengender Check den gesamten Readiness-Endpunkt blockiert
|
||||
// (Akzeptanzkriterium 2 / Pruefung 2: Antwort innerhalb definierter Zeit).
|
||||
const DefaultCheckTimeout = 2 * time.Second
|
||||
|
||||
// Registry haelt alle benannten Checks eines Dienstes.
|
||||
type Registry struct {
|
||||
checks map[string]Checker
|
||||
timeout time.Duration
|
||||
}
|
||||
|
||||
func NewRegistry() *Registry {
|
||||
return &Registry{checks: make(map[string]Checker), timeout: DefaultCheckTimeout}
|
||||
}
|
||||
|
||||
func (r *Registry) WithTimeout(d time.Duration) *Registry {
|
||||
return &Registry{checks: r.checks, timeout: d}
|
||||
}
|
||||
|
||||
// Register fuegt einen benannten Check hinzu (z.B. "database", "queue").
|
||||
func (r *Registry) Register(name string, c Checker) {
|
||||
r.checks[name] = c
|
||||
}
|
||||
|
||||
// Result ist der Ausgang eines einzelnen Checks.
|
||||
type Result struct {
|
||||
OK bool
|
||||
Error string
|
||||
}
|
||||
|
||||
// CheckAll fuehrt alle registrierten Checks NEBENLAEUFIG mit je eigenem
|
||||
// Timeout aus (Akzeptanzkriterium 1: echte Abhaengigkeiten statt Prozess-
|
||||
// status) und liefert ready=false, sobald irgendein Check fehlschlaegt
|
||||
// (Akzeptanzkriterium 3: ein Ausfall wird sichtbar).
|
||||
func (r *Registry) CheckAll(ctx context.Context) (ready bool, results map[string]Result) {
|
||||
type namedResult struct {
|
||||
name string
|
||||
result Result
|
||||
}
|
||||
ch := make(chan namedResult, len(r.checks))
|
||||
|
||||
for name, checker := range r.checks {
|
||||
go func(name string, checker Checker) {
|
||||
checkCtx, cancel := context.WithTimeout(ctx, r.timeout)
|
||||
defer cancel()
|
||||
err := checker.Check(checkCtx)
|
||||
if err != nil {
|
||||
ch <- namedResult{name, Result{OK: false, Error: err.Error()}}
|
||||
return
|
||||
}
|
||||
ch <- namedResult{name, Result{OK: true}}
|
||||
}(name, checker)
|
||||
}
|
||||
|
||||
results = make(map[string]Result, len(r.checks))
|
||||
ready = true
|
||||
for i := 0; i < len(r.checks); i++ {
|
||||
nr := <-ch
|
||||
results[nr.name] = nr.result
|
||||
if !nr.result.OK {
|
||||
ready = false
|
||||
}
|
||||
}
|
||||
return ready, results
|
||||
}
|
||||
@@ -1,161 +0,0 @@
|
||||
package health
|
||||
|
||||
import (
|
||||
"context"
|
||||
"encoding/json"
|
||||
"errors"
|
||||
"net/http"
|
||||
"net/http/httptest"
|
||||
"os"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
// Akzeptanzkriterium 1 + Pruefung 1: simulierter Datenbankausfall fuehrt zu
|
||||
// "nicht bereit".
|
||||
func TestReadinessHandler_ReportsNotReadyOnDatabaseFailure(t *testing.T) {
|
||||
adminDSN := os.Getenv("TEST_ADMIN_DSN")
|
||||
if adminDSN == "" {
|
||||
t.Skip("TEST_ADMIN_DSN nicht gesetzt, Integrationstest uebersprungen")
|
||||
}
|
||||
ctx := context.Background()
|
||||
|
||||
pool, err := pgxpool.New(ctx, adminDSN)
|
||||
if err != nil {
|
||||
t.Fatalf("pool: %v", err)
|
||||
}
|
||||
// Datenbankausfall simulieren: Pool sofort schliessen, bevor der Check laeuft.
|
||||
pool.Close()
|
||||
|
||||
reg := NewRegistry()
|
||||
reg.Register("database", DatabaseChecker(pool))
|
||||
|
||||
req := httptest.NewRequest(http.MethodGet, "/readyz", nil)
|
||||
rec := httptest.NewRecorder()
|
||||
reg.ReadinessHandler()(rec, req)
|
||||
|
||||
if rec.Code != http.StatusServiceUnavailable {
|
||||
t.Fatalf("status = %d, want 503 bei db-ausfall", rec.Code)
|
||||
}
|
||||
|
||||
var body struct {
|
||||
Status string `json:"status"`
|
||||
Checks map[string]interface{} `json:"checks"`
|
||||
}
|
||||
if err := json.Unmarshal(rec.Body.Bytes(), &body); err != nil {
|
||||
t.Fatalf("body parsen: %v", err)
|
||||
}
|
||||
if body.Status != "not_ready" {
|
||||
t.Fatalf("status-feld = %q, want not_ready", body.Status)
|
||||
}
|
||||
if _, ok := body.Checks["database"]; !ok {
|
||||
t.Fatal("erwartet 'database' im checks-ergebnis")
|
||||
}
|
||||
}
|
||||
|
||||
func TestReadinessHandler_ReportsReadyWhenAllChecksPass(t *testing.T) {
|
||||
adminDSN := os.Getenv("TEST_ADMIN_DSN")
|
||||
if adminDSN == "" {
|
||||
t.Skip("TEST_ADMIN_DSN nicht gesetzt, Integrationstest uebersprungen")
|
||||
}
|
||||
ctx := context.Background()
|
||||
|
||||
pool, err := pgxpool.New(ctx, adminDSN)
|
||||
if err != nil {
|
||||
t.Fatalf("pool: %v", err)
|
||||
}
|
||||
defer pool.Close()
|
||||
|
||||
reg := NewRegistry()
|
||||
reg.Register("database", DatabaseChecker(pool))
|
||||
reg.Register("queue", QueueChecker(pool))
|
||||
|
||||
req := httptest.NewRequest(http.MethodGet, "/readyz", nil)
|
||||
rec := httptest.NewRecorder()
|
||||
reg.ReadinessHandler()(rec, req)
|
||||
|
||||
if rec.Code != http.StatusOK {
|
||||
t.Fatalf("status = %d, want 200 bei funktionierenden abhaengigkeiten", rec.Code)
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 + Pruefung 3: Liveness und Readiness unterscheiden
|
||||
// sich nachweislich im Fehlerfall.
|
||||
func TestLivenessAndReadiness_DifferOnDatabaseFailure(t *testing.T) {
|
||||
adminDSN := os.Getenv("TEST_ADMIN_DSN")
|
||||
if adminDSN == "" {
|
||||
t.Skip("TEST_ADMIN_DSN nicht gesetzt, Integrationstest uebersprungen")
|
||||
}
|
||||
ctx := context.Background()
|
||||
|
||||
pool, err := pgxpool.New(ctx, adminDSN)
|
||||
if err != nil {
|
||||
t.Fatalf("pool: %v", err)
|
||||
}
|
||||
pool.Close() // db-ausfall simulieren
|
||||
|
||||
reg := NewRegistry()
|
||||
reg.Register("database", DatabaseChecker(pool))
|
||||
|
||||
livenessRec := httptest.NewRecorder()
|
||||
LivenessHandler(livenessRec, httptest.NewRequest(http.MethodGet, "/livez", nil))
|
||||
if livenessRec.Code != http.StatusOK {
|
||||
t.Fatalf("liveness status = %d, want 200 trotz db-ausfall (liveness prueft keine abhaengigkeiten)", livenessRec.Code)
|
||||
}
|
||||
|
||||
readinessRec := httptest.NewRecorder()
|
||||
reg.ReadinessHandler()(readinessRec, httptest.NewRequest(http.MethodGet, "/readyz", nil))
|
||||
if readinessRec.Code != http.StatusServiceUnavailable {
|
||||
t.Fatalf("readiness status = %d, want 503 bei db-ausfall", readinessRec.Code)
|
||||
}
|
||||
|
||||
if livenessRec.Code == readinessRec.Code {
|
||||
t.Fatal("liveness und readiness sollten sich im db-ausfall-fall unterscheiden")
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 + Pruefung 2: Health-Endpunkt antwortet auch bei
|
||||
// haengendem Check innerhalb definierter Zeit (Timeout begrenzt die Dauer).
|
||||
func TestReadinessHandler_RespondsWithinTimeoutEvenWithHangingCheck(t *testing.T) {
|
||||
reg := NewRegistry().WithTimeout(50 * time.Millisecond)
|
||||
reg.Register("haengender_dienst", CheckerFunc(func(ctx context.Context) error {
|
||||
select {
|
||||
case <-time.After(10 * time.Second): // wuerde ohne timeout ewig blockieren
|
||||
return nil
|
||||
case <-ctx.Done():
|
||||
return ctx.Err()
|
||||
}
|
||||
}))
|
||||
|
||||
start := time.Now()
|
||||
req := httptest.NewRequest(http.MethodGet, "/readyz", nil)
|
||||
rec := httptest.NewRecorder()
|
||||
reg.ReadinessHandler()(rec, req)
|
||||
elapsed := time.Since(start)
|
||||
|
||||
if elapsed > time.Second {
|
||||
t.Fatalf("readiness handler brauchte %s, erwartet deutlich unter 1s durch timeout", elapsed)
|
||||
}
|
||||
if rec.Code != http.StatusServiceUnavailable {
|
||||
t.Fatalf("status = %d, want 503 fuer haengenden/timeout-check", rec.Code)
|
||||
}
|
||||
}
|
||||
|
||||
func TestCheckAll_MultipleChecksRunConcurrently(t *testing.T) {
|
||||
reg := NewRegistry().WithTimeout(time.Second)
|
||||
reg.Register("a", CheckerFunc(func(ctx context.Context) error { return nil }))
|
||||
reg.Register("b", CheckerFunc(func(ctx context.Context) error { return errors.New("kaputt") }))
|
||||
|
||||
ready, results := reg.CheckAll(context.Background())
|
||||
if ready {
|
||||
t.Fatal("erwartet ready=false, da 'b' fehlschlaegt")
|
||||
}
|
||||
if !results["a"].OK {
|
||||
t.Fatalf("erwartet 'a' ok, habe %+v", results["a"])
|
||||
}
|
||||
if results["b"].OK || results["b"].Error == "" {
|
||||
t.Fatalf("erwartet 'b' fehlgeschlagen mit fehlertext, habe %+v", results["b"])
|
||||
}
|
||||
}
|
||||
@@ -1,19 +0,0 @@
|
||||
package metrics
|
||||
|
||||
import "github.com/prometheus/client_golang/prometheus"
|
||||
|
||||
// NewCoreRegistry liefert das Prometheus-Registry fuer die EIGENEN
|
||||
// Kennzahlen des Core-Dienstes (Akzeptanzkriterium 1) — alle Namen tragen
|
||||
// das Praefix "nexarch_core_" gemaess der im Paketkommentar dokumentierten
|
||||
// Namenskonvention (Akzeptanzkriterium 3). Ein eigenes Registry statt des
|
||||
// globalen DefaultRegisterer, damit Tests unabhaengig voneinander sind.
|
||||
func NewCoreRegistry() *prometheus.Registry {
|
||||
reg := prometheus.NewRegistry()
|
||||
reg.MustRegister(
|
||||
prometheus.NewGaugeFunc(prometheus.GaugeOpts{
|
||||
Name: "nexarch_core_up",
|
||||
Help: "1, solange der Core-Dienst laeuft und Metriken liefern kann.",
|
||||
}, func() float64 { return 1 }),
|
||||
)
|
||||
return reg
|
||||
}
|
||||
@@ -1,168 +0,0 @@
|
||||
// Package metrics implementiert Core OPS-03: einen zentralen /metrics-
|
||||
// Endpunkt im Prometheus-Textformat, der Kennzahlen des Core-Dienstes UND
|
||||
// aggregierte Kennzahlen aller registrierten Module bereitstellt — offenes
|
||||
// Pull-Modell nach Prometheus-Vorbild, kein proprietaerer Push-Mechanismus.
|
||||
//
|
||||
// Namenskonvention (Akzeptanzkriterium 3, modulübergreifend konsistent):
|
||||
//
|
||||
// nexarch_core_<name> — Kennzahlen des Core-Dienstes selbst
|
||||
// nexarch_module_<modul>_<name> — von einem Modul gescrapte Kennzahl
|
||||
// <name>, umbenannt mit dem
|
||||
// Modulnamen als Praefix
|
||||
//
|
||||
// Ein Modul liefert seine eigenen Kennzahlen unter EIGENEM Namen (z.B.
|
||||
// "requests_total") unter seinem eigenen /metrics-Endpunkt — dieses Paket
|
||||
// benennt sie beim Einsammeln konsistent um, damit im aggregierten Core-
|
||||
// Endpunkt niemals zwei Module denselben Metrik-Namen kollidieren lassen.
|
||||
package metrics
|
||||
|
||||
import (
|
||||
"context"
|
||||
"fmt"
|
||||
"net/http"
|
||||
"time"
|
||||
|
||||
"github.com/prometheus/client_golang/prometheus"
|
||||
dto "github.com/prometheus/client_model/go"
|
||||
"github.com/prometheus/common/expfmt"
|
||||
"github.com/prometheus/common/model"
|
||||
)
|
||||
|
||||
// init erzwingt das klassische Prometheus-Namensschema (a-z, A-Z, 0-9, _)
|
||||
// fuer die Namensvalidierung von expfmt/model — ohne diese explizite
|
||||
// Festlegung liefert die Bibliothek "Invalid name validation scheme
|
||||
// requested: unset" beim Parsen/Kodieren, da sie den globalen Default in
|
||||
// dieser Version nicht mehr implizit setzt.
|
||||
func init() {
|
||||
model.NameValidationScheme = model.LegacyValidation
|
||||
}
|
||||
|
||||
// Source ist EIN registriertes Modul mit seinem eigenen /metrics-Endpunkt
|
||||
// (siehe internal/health fuer das analoge Muster bei Readiness-Checks).
|
||||
type Source struct {
|
||||
ModuleName string
|
||||
MetricsURL string
|
||||
}
|
||||
|
||||
// SourceProvider liefert die aktuell registrierten Module — typischerweise
|
||||
// rueckgebunden an internal/moduleregistry.Registry.List (API-02) ueber
|
||||
// einen kleinen Adapter im aufrufenden Code, damit dieses Paket
|
||||
// internal/moduleregistry nicht direkt importieren muss (Kein Umbau
|
||||
// angrenzender Bereiche). Ein NEU registriertes Modul erscheint automatisch
|
||||
// beim naechsten Aufruf von Aggregator.Handler, OHNE Codeaenderung an diesem
|
||||
// Paket (Akzeptanzkriterium 2 / Pruefung 2).
|
||||
type SourceProvider func(ctx context.Context) ([]Source, error)
|
||||
|
||||
// FetchTimeout begrenzt, wie lange EIN Modul-Scrape maximal dauern darf —
|
||||
// ein haengendes Modul darf den gesamten Aggregations-Request nicht
|
||||
// verzoegern (Pruefung 1: Antwort unter Last innerhalb definierter Zeit).
|
||||
const FetchTimeout = 2 * time.Second
|
||||
|
||||
// Aggregator sammelt Core-eigene Metriken (coreGatherer) und die Metriken
|
||||
// aller ueber sourceProvider gemeldeten Module in EINER Antwort ein.
|
||||
type Aggregator struct {
|
||||
coreGatherer prometheus.Gatherer
|
||||
sourceProvider SourceProvider
|
||||
client *http.Client
|
||||
}
|
||||
|
||||
func NewAggregator(coreGatherer prometheus.Gatherer, sourceProvider SourceProvider) *Aggregator {
|
||||
return &Aggregator{
|
||||
coreGatherer: coreGatherer,
|
||||
sourceProvider: sourceProvider,
|
||||
client: &http.Client{Timeout: FetchTimeout},
|
||||
}
|
||||
}
|
||||
|
||||
// Gather implementiert prometheus.Gatherer: liefert Core-Metriken PLUS alle
|
||||
// erreichbaren Modul-Metriken (umbenannt gemaess Namenskonvention) in einer
|
||||
// gemeinsamen Liste von MetricFamilies.
|
||||
func (a *Aggregator) Gather(ctx context.Context) ([]*dto.MetricFamily, error) {
|
||||
families, err := a.coreGatherer.Gather()
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("core-metriken einsammeln: %w", err)
|
||||
}
|
||||
|
||||
sources, err := a.sourceProvider(ctx)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("modul-quellen ermitteln: %w", err)
|
||||
}
|
||||
|
||||
// Module werden NEBENLAEUFIG gescrapt (dasselbe Muster wie
|
||||
// internal/health.Registry.CheckAll) — ein langsames/nicht erreichbares
|
||||
// Modul haelt weder andere Module noch den Gesamt-Request auf.
|
||||
type fetchResult struct {
|
||||
families []*dto.MetricFamily
|
||||
}
|
||||
resultCh := make(chan fetchResult, len(sources))
|
||||
for _, src := range sources {
|
||||
go func(src Source) {
|
||||
fetchCtx, cancel := context.WithTimeout(ctx, FetchTimeout)
|
||||
defer cancel()
|
||||
mf, err := a.fetchAndRename(fetchCtx, src)
|
||||
if err != nil {
|
||||
resultCh <- fetchResult{} // Fehlerfall: einfach nichts beitragen, Aggregation laeuft weiter
|
||||
return
|
||||
}
|
||||
resultCh <- fetchResult{families: mf}
|
||||
}(src)
|
||||
}
|
||||
for range sources {
|
||||
r := <-resultCh
|
||||
families = append(families, r.families...)
|
||||
}
|
||||
|
||||
return families, nil
|
||||
}
|
||||
|
||||
// fetchAndRename ruft die /metrics-URL eines Moduls ab, parst das
|
||||
// Prometheus-Textformat und benennt jede Metrik gemaess der
|
||||
// Namenskonvention um (Akzeptanzkriterium 2 + 3).
|
||||
func (a *Aggregator) fetchAndRename(ctx context.Context, src Source) ([]*dto.MetricFamily, error) {
|
||||
req, err := http.NewRequestWithContext(ctx, http.MethodGet, src.MetricsURL, nil)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
resp, err := a.client.Do(req)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
defer resp.Body.Close()
|
||||
if resp.StatusCode != http.StatusOK {
|
||||
return nil, fmt.Errorf("modul %s: unerwarteter status %d", src.ModuleName, resp.StatusCode)
|
||||
}
|
||||
|
||||
parser := expfmt.NewTextParser(model.LegacyValidation)
|
||||
parsed, err := parser.TextToMetricFamilies(resp.Body)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("modul %s: metrik-text nicht parsebar: %w", src.ModuleName, err)
|
||||
}
|
||||
|
||||
out := make([]*dto.MetricFamily, 0, len(parsed))
|
||||
for name, mf := range parsed {
|
||||
renamed := fmt.Sprintf("nexarch_module_%s_%s", src.ModuleName, name)
|
||||
mf.Name = &renamed
|
||||
out = append(out, mf)
|
||||
}
|
||||
return out, nil
|
||||
}
|
||||
|
||||
// Handler liefert einen HTTP-Handler, der Gather aufruft und das Ergebnis im
|
||||
// Prometheus-Textformat ausgibt (Akzeptanzkriterium 1).
|
||||
func (a *Aggregator) Handler() http.HandlerFunc {
|
||||
return func(w http.ResponseWriter, r *http.Request) {
|
||||
families, err := a.Gather(r.Context())
|
||||
if err != nil {
|
||||
http.Error(w, "metriken konnten nicht eingesammelt werden", http.StatusInternalServerError)
|
||||
return
|
||||
}
|
||||
|
||||
w.Header().Set("Content-Type", string(expfmt.NewFormat(expfmt.TypeTextPlain)))
|
||||
enc := expfmt.NewEncoder(w, expfmt.NewFormat(expfmt.TypeTextPlain))
|
||||
for _, mf := range families {
|
||||
if err := enc.Encode(mf); err != nil {
|
||||
return
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -1,179 +0,0 @@
|
||||
package metrics
|
||||
|
||||
import (
|
||||
"context"
|
||||
"fmt"
|
||||
"net/http"
|
||||
"net/http/httptest"
|
||||
"strings"
|
||||
"sync"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/prometheus/common/expfmt"
|
||||
"github.com/prometheus/common/model"
|
||||
)
|
||||
|
||||
func fakeModuleServer(metricName string) *httptest.Server {
|
||||
return httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
|
||||
w.Header().Set("Content-Type", "text/plain; version=0.0.4")
|
||||
fmt.Fprintf(w, "# HELP %s ein test-zaehler\n# TYPE %s counter\n%s 42\n", metricName, metricName, metricName)
|
||||
}))
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 1: Core liefert unter dem Handler valides
|
||||
// Prometheus-Textformat mit den eigenen Kennzahlen.
|
||||
func TestHandler_ServesCoreMetricsInPrometheusFormat(t *testing.T) {
|
||||
agg := NewAggregator(NewCoreRegistry(), func(ctx context.Context) ([]Source, error) { return nil, nil })
|
||||
|
||||
req := httptest.NewRequest(http.MethodGet, "/metrics", nil)
|
||||
rec := httptest.NewRecorder()
|
||||
agg.Handler()(rec, req)
|
||||
|
||||
if rec.Code != http.StatusOK {
|
||||
t.Fatalf("status = %d, want 200", rec.Code)
|
||||
}
|
||||
|
||||
parser := expfmt.NewTextParser(model.LegacyValidation)
|
||||
families, err := parser.TextToMetricFamilies(strings.NewReader(rec.Body.String()))
|
||||
if err != nil {
|
||||
t.Fatalf("antwort ist kein valides prometheus-textformat: %v", err)
|
||||
}
|
||||
if _, ok := families["nexarch_core_up"]; !ok {
|
||||
t.Fatalf("erwartet 'nexarch_core_up' unter den core-metriken, habe: %v", keysOf(families))
|
||||
}
|
||||
}
|
||||
|
||||
func keysOf[V any](m map[string]V) []string {
|
||||
out := make([]string, 0, len(m))
|
||||
for k := range m {
|
||||
out = append(out, k)
|
||||
}
|
||||
return out
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 + Pruefung 2: ein NEU registriertes Modul erscheint
|
||||
// in der Aggregation, OHNE dass dieses Paket oder der Aufrufer Code
|
||||
// aendern muss — die Quelle kommt ausschliesslich aus sourceProvider.
|
||||
func TestHandler_NewlyRegisteredModuleAppearsWithoutCodeChange(t *testing.T) {
|
||||
moduleServer := fakeModuleServer("requests_total")
|
||||
defer moduleServer.Close()
|
||||
|
||||
// Simuliert eine sich zur Laufzeit aendernde Modul-Liste (z.B. aus
|
||||
// SourceStore.Provide) — zunaechst LEER, dann mit einem Eintrag.
|
||||
var sources []Source
|
||||
var mu sync.Mutex
|
||||
provider := func(ctx context.Context) ([]Source, error) {
|
||||
mu.Lock()
|
||||
defer mu.Unlock()
|
||||
out := make([]Source, len(sources))
|
||||
copy(out, sources)
|
||||
return out, nil
|
||||
}
|
||||
|
||||
agg := NewAggregator(NewCoreRegistry(), provider)
|
||||
|
||||
// Vor der Registrierung: Modul-Metrik nicht vorhanden.
|
||||
rec1 := httptest.NewRecorder()
|
||||
agg.Handler()(rec1, httptest.NewRequest(http.MethodGet, "/metrics", nil))
|
||||
if strings.Contains(rec1.Body.String(), "requests_total") {
|
||||
t.Fatal("modul-metrik haette vor registrierung nicht erscheinen duerfen")
|
||||
}
|
||||
|
||||
// Modul wird "registriert" (kein Code hier oder in metrics.go aendert sich).
|
||||
mu.Lock()
|
||||
sources = append(sources, Source{ModuleName: "dms", MetricsURL: moduleServer.URL})
|
||||
mu.Unlock()
|
||||
|
||||
rec2 := httptest.NewRecorder()
|
||||
agg.Handler()(rec2, httptest.NewRequest(http.MethodGet, "/metrics", nil))
|
||||
body := rec2.Body.String()
|
||||
if !strings.Contains(body, "nexarch_module_dms_requests_total") {
|
||||
t.Fatalf("erwartet umbenannte modul-metrik 'nexarch_module_dms_requests_total' nach registrierung, body:\n%s", body)
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 3: Namenskonvention "nexarch_module_<modul>_<name>"
|
||||
// wird tatsaechlich angewendet.
|
||||
func TestFetchAndRename_AppliesNamingConvention(t *testing.T) {
|
||||
moduleServer := fakeModuleServer("queue_depth")
|
||||
defer moduleServer.Close()
|
||||
|
||||
agg := NewAggregator(NewCoreRegistry(), nil)
|
||||
families, err := agg.fetchAndRename(context.Background(), Source{ModuleName: "mail", MetricsURL: moduleServer.URL})
|
||||
if err != nil {
|
||||
t.Fatalf("fetchAndRename: %v", err)
|
||||
}
|
||||
if len(families) != 1 || families[0].GetName() != "nexarch_module_mail_queue_depth" {
|
||||
t.Fatalf("erwartet genau 1 metrik 'nexarch_module_mail_queue_depth', habe: %+v", families)
|
||||
}
|
||||
}
|
||||
|
||||
// Ein nicht erreichbares Modul darf die Aggregation der uebrigen und die
|
||||
// Gesamtantwort nicht verhindern (dieselbe Resilienz wie OPS-02).
|
||||
func TestHandler_UnreachableModuleDoesNotBreakAggregation(t *testing.T) {
|
||||
reachable := fakeModuleServer("healthy_metric")
|
||||
defer reachable.Close()
|
||||
unreachable := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {}))
|
||||
unreachableURL := unreachable.URL
|
||||
unreachable.Close() // sofort schliessen -> Verbindung schlaegt fehl
|
||||
|
||||
provider := func(ctx context.Context) ([]Source, error) {
|
||||
return []Source{
|
||||
{ModuleName: "ok", MetricsURL: reachable.URL},
|
||||
{ModuleName: "kaputt", MetricsURL: unreachableURL},
|
||||
}, nil
|
||||
}
|
||||
agg := NewAggregator(NewCoreRegistry(), provider)
|
||||
|
||||
rec := httptest.NewRecorder()
|
||||
agg.Handler()(rec, httptest.NewRequest(http.MethodGet, "/metrics", nil))
|
||||
|
||||
if rec.Code != http.StatusOK {
|
||||
t.Fatalf("status = %d, want 200 trotz einem nicht erreichbaren modul", rec.Code)
|
||||
}
|
||||
body := rec.Body.String()
|
||||
if !strings.Contains(body, "nexarch_module_ok_healthy_metric") {
|
||||
t.Fatal("erreichbares modul haette trotz ausfall des anderen aggregiert werden sollen")
|
||||
}
|
||||
if strings.Contains(body, "kaputt") {
|
||||
t.Fatal("nicht erreichbares modul haette keine metrik beitragen duerfen")
|
||||
}
|
||||
}
|
||||
|
||||
// Pruefung 1: Endpunkt antwortet unter mehreren gleichzeitigen Anfragen
|
||||
// innerhalb definierter Zeit — kein unbeschraenktes Blockieren durch
|
||||
// langsame Module (FetchTimeout begrenzt jeden Scrape).
|
||||
func TestHandler_RespondsWithinBoundedTimeUnderLoad(t *testing.T) {
|
||||
hangingServer := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
|
||||
time.Sleep(10 * time.Second) // wuerde ohne timeout jede anfrage blockieren
|
||||
}))
|
||||
defer hangingServer.Close()
|
||||
|
||||
provider := func(ctx context.Context) ([]Source, error) {
|
||||
return []Source{{ModuleName: "haengend", MetricsURL: hangingServer.URL}}, nil
|
||||
}
|
||||
agg := NewAggregator(NewCoreRegistry(), provider)
|
||||
|
||||
const concurrentRequests = 10
|
||||
var wg sync.WaitGroup
|
||||
start := time.Now()
|
||||
for i := 0; i < concurrentRequests; i++ {
|
||||
wg.Add(1)
|
||||
go func() {
|
||||
defer wg.Done()
|
||||
rec := httptest.NewRecorder()
|
||||
agg.Handler()(rec, httptest.NewRequest(http.MethodGet, "/metrics", nil))
|
||||
if rec.Code != http.StatusOK {
|
||||
t.Errorf("status = %d, want 200", rec.Code)
|
||||
}
|
||||
}()
|
||||
}
|
||||
wg.Wait()
|
||||
elapsed := time.Since(start)
|
||||
|
||||
if elapsed > FetchTimeout+3*time.Second {
|
||||
t.Fatalf("%d gleichzeitige anfragen brauchten %s, erwartet deutlich unter %s durch FetchTimeout",
|
||||
concurrentRequests, elapsed, FetchTimeout+3*time.Second)
|
||||
}
|
||||
}
|
||||
@@ -1,52 +0,0 @@
|
||||
package metrics
|
||||
|
||||
import (
|
||||
"context"
|
||||
"fmt"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
// SourceStore persistiert, welche Module ihre Metriken unter welcher URL
|
||||
// bereitstellen — dieselbe Postgres-basierte "kein Code-Deploy noetig"-
|
||||
// Konvention wie internal/statuspage.Store.RegisterTarget (OPS-02): ein neu
|
||||
// registriertes Modul erscheint automatisch in der Aggregation, sobald es
|
||||
// hier eingetragen ist (Akzeptanzkriterium 2 / Pruefung 2).
|
||||
type SourceStore struct {
|
||||
pool *pgxpool.Pool
|
||||
}
|
||||
|
||||
func NewSourceStore(pool *pgxpool.Pool) *SourceStore {
|
||||
return &SourceStore{pool: pool}
|
||||
}
|
||||
|
||||
func (s *SourceStore) RegisterSource(ctx context.Context, moduleName, metricsURL string) error {
|
||||
_, err := s.pool.Exec(ctx, `
|
||||
INSERT INTO metrics_sources (module_name, metrics_url)
|
||||
VALUES ($1, $2)
|
||||
ON CONFLICT (module_name) DO UPDATE SET metrics_url = $2
|
||||
`, moduleName, metricsURL)
|
||||
if err != nil {
|
||||
return fmt.Errorf("metrik-quelle speichern: %w", err)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
// Provide implementiert SourceProvider direkt aus der Datenbank.
|
||||
func (s *SourceStore) Provide(ctx context.Context) ([]Source, error) {
|
||||
rows, err := s.pool.Query(ctx, `SELECT module_name, metrics_url FROM metrics_sources ORDER BY module_name`)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("metrik-quellen auflisten: %w", err)
|
||||
}
|
||||
defer rows.Close()
|
||||
|
||||
var out []Source
|
||||
for rows.Next() {
|
||||
var src Source
|
||||
if err := rows.Scan(&src.ModuleName, &src.MetricsURL); err != nil {
|
||||
return nil, fmt.Errorf("metrik-quelle lesen: %w", err)
|
||||
}
|
||||
out = append(out, src)
|
||||
}
|
||||
return out, rows.Err()
|
||||
}
|
||||
@@ -1,60 +0,0 @@
|
||||
package metrics
|
||||
|
||||
import (
|
||||
"context"
|
||||
"fmt"
|
||||
"os"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
func setupSourceStoreTest(t *testing.T) (*SourceStore, func()) {
|
||||
t.Helper()
|
||||
adminDSN := os.Getenv("TEST_ADMIN_DSN")
|
||||
if adminDSN == "" {
|
||||
t.Skip("TEST_ADMIN_DSN nicht gesetzt, Integrationstest uebersprungen")
|
||||
}
|
||||
ctx := context.Background()
|
||||
pool, err := pgxpool.New(ctx, adminDSN)
|
||||
if err != nil {
|
||||
t.Fatalf("pool: %v", err)
|
||||
}
|
||||
if _, err := pool.Exec(ctx, `
|
||||
CREATE TABLE IF NOT EXISTS metrics_sources (module_name TEXT PRIMARY KEY, metrics_url TEXT NOT NULL)
|
||||
`); err != nil {
|
||||
t.Fatalf("schema: %v", err)
|
||||
}
|
||||
cleanup := func() { pool.Close() }
|
||||
return NewSourceStore(pool), cleanup
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 / Pruefung 2 auf Persistenz-Ebene: eine ueber die
|
||||
// Datenbank registrierte Quelle ist sofort ueber Provide() sichtbar — genau
|
||||
// der Mechanismus, der ein neues Modul ohne Core-Codeaenderung erscheinen
|
||||
// laesst.
|
||||
func TestSourceStore_RegisterSourceAppearsInProvide(t *testing.T) {
|
||||
store, cleanup := setupSourceStoreTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
name := fmt.Sprintf("modul-%d", time.Now().UnixNano())
|
||||
|
||||
if err := store.RegisterSource(ctx, name, "http://example.invalid/metrics"); err != nil {
|
||||
t.Fatalf("registersource: %v", err)
|
||||
}
|
||||
|
||||
sources, err := store.Provide(ctx)
|
||||
if err != nil {
|
||||
t.Fatalf("provide: %v", err)
|
||||
}
|
||||
found := false
|
||||
for _, s := range sources {
|
||||
if s.ModuleName == name {
|
||||
found = true
|
||||
}
|
||||
}
|
||||
if !found {
|
||||
t.Fatalf("erwartet %s in provide()-ergebnis, habe: %+v", name, sources)
|
||||
}
|
||||
}
|
||||
@@ -1,99 +0,0 @@
|
||||
package moduleregistry
|
||||
|
||||
import (
|
||||
"context"
|
||||
"crypto/rand"
|
||||
"crypto/sha256"
|
||||
"crypto/subtle"
|
||||
"encoding/hex"
|
||||
"errors"
|
||||
"fmt"
|
||||
|
||||
"github.com/jackc/pgx/v5"
|
||||
)
|
||||
|
||||
var (
|
||||
ErrModuleNotRegistered = errors.New("moduleregistry: modul muss vor provisionierung registriert sein")
|
||||
ErrInvalidCredential = errors.New("moduleregistry: ungueltiges oder fehlendes service-credential")
|
||||
)
|
||||
|
||||
// Provision stellt ein Service-Credential (Client-ID + Secret) fuer eine
|
||||
// Modul-Instanz aus (Akzeptanzkriterium 4). Das Secret wird NUR beim
|
||||
// Ausstellen im Klartext zurueckgegeben, gespeichert wird ausschliesslich
|
||||
// dessen SHA-256-Hash.
|
||||
func (r *Registry) Provision(ctx context.Context, moduleName string) (clientID, secret string, err error) {
|
||||
if _, err := r.Get(ctx, moduleName); err != nil {
|
||||
if errors.Is(err, ErrModuleNotFound) {
|
||||
return "", "", ErrModuleNotRegistered
|
||||
}
|
||||
return "", "", err
|
||||
}
|
||||
|
||||
clientID, err = randomToken(16)
|
||||
if err != nil {
|
||||
return "", "", fmt.Errorf("client-id erzeugen: %w", err)
|
||||
}
|
||||
secret, err = randomToken(32)
|
||||
if err != nil {
|
||||
return "", "", fmt.Errorf("secret erzeugen: %w", err)
|
||||
}
|
||||
hash := hashSecret(secret)
|
||||
|
||||
_, err = r.pool.Exec(ctx, `
|
||||
INSERT INTO module_credentials (module_name, client_id, secret_hash, issued_at)
|
||||
VALUES ($1, $2, $3, now())
|
||||
ON CONFLICT (module_name) DO UPDATE SET client_id = $2, secret_hash = $3, issued_at = now()
|
||||
`, moduleName, clientID, hash)
|
||||
if err != nil {
|
||||
return "", "", fmt.Errorf("credential speichern: %w", err)
|
||||
}
|
||||
return clientID, secret, nil
|
||||
}
|
||||
|
||||
// Authenticate prueft ein Service-Credential timing-safe (Referenzmuster
|
||||
// siehe AUD-02) — Aufrufe ohne gueltiges Credential werden abgelehnt
|
||||
// (Akzeptanzkriterium 4 / Pruefung 4).
|
||||
func (r *Registry) Authenticate(ctx context.Context, clientID, secret string) (moduleName string, ok bool, err error) {
|
||||
if clientID == "" || secret == "" {
|
||||
return "", false, nil
|
||||
}
|
||||
|
||||
var storedHash []byte
|
||||
err = r.pool.QueryRow(ctx, `
|
||||
SELECT module_name, secret_hash FROM module_credentials WHERE client_id = $1
|
||||
`, clientID).Scan(&moduleName, &storedHash)
|
||||
if err != nil {
|
||||
if errors.Is(err, pgx.ErrNoRows) {
|
||||
return "", false, nil
|
||||
}
|
||||
return "", false, fmt.Errorf("credential lesen: %w", err)
|
||||
}
|
||||
|
||||
if !timingSafeEqual(hashSecret(secret), storedHash) {
|
||||
return "", false, nil
|
||||
}
|
||||
return moduleName, true, nil
|
||||
}
|
||||
|
||||
func randomToken(n int) (string, error) {
|
||||
buf := make([]byte, n)
|
||||
if _, err := rand.Read(buf); err != nil {
|
||||
return "", err
|
||||
}
|
||||
return hex.EncodeToString(buf), nil
|
||||
}
|
||||
|
||||
func hashSecret(secret string) []byte {
|
||||
sum := sha256.Sum256([]byte(secret))
|
||||
return sum[:]
|
||||
}
|
||||
|
||||
// timingSafeEqual folgt derselben Referenzimplementierung wie AUD-02
|
||||
// (subtle.ConstantTimeCompare) — projektweite Konvention fuer jeden
|
||||
// sicherheitsrelevanten Vergleich.
|
||||
func timingSafeEqual(a, b []byte) bool {
|
||||
if len(a) != len(b) {
|
||||
return false
|
||||
}
|
||||
return subtle.ConstantTimeCompare(a, b) == 1
|
||||
}
|
||||
@@ -1,46 +0,0 @@
|
||||
package moduleregistry
|
||||
|
||||
import "net/http"
|
||||
|
||||
// RequireActiveModule weist Anfragen an ein nicht aktiviertes Modul ZENTRAL
|
||||
// ab, bevor der eigentliche Modul-Handler erreicht wird (Akzeptanzkriterium 2 /
|
||||
// Pruefung 1) — Casbin-Prinzip: Durchsetzung als Middleware statt verstreuter
|
||||
// Pruefungen in jedem Handler. tenantSlug/moduleName werden hier ueber
|
||||
// Query-Parameter gelesen (echte Extraktion aus JWT/Tenant-Kontext ist
|
||||
// API-05/TEN-06, nicht Teil dieser Kachel).
|
||||
func (r *Registry) RequireActiveModule(moduleName string, next http.HandlerFunc) http.HandlerFunc {
|
||||
return func(w http.ResponseWriter, req *http.Request) {
|
||||
tenantSlug := req.URL.Query().Get("tenant")
|
||||
active, err := r.IsActive(req.Context(), tenantSlug, moduleName)
|
||||
if err != nil {
|
||||
http.Error(w, "aktivierungspruefung fehlgeschlagen", http.StatusInternalServerError)
|
||||
return
|
||||
}
|
||||
if !active {
|
||||
http.Error(w, "modul nicht aktiviert", http.StatusForbidden)
|
||||
return
|
||||
}
|
||||
next(w, req)
|
||||
}
|
||||
}
|
||||
|
||||
// RequireServiceCredential authentifiziert eine Modul-Instanz ueber ihr
|
||||
// Service-Credential (X-Client-Id/X-Client-Secret-Header) BEVOR der
|
||||
// eigentliche Handler erreicht wird (Akzeptanzkriterium 4 / Pruefung 4).
|
||||
func (r *Registry) RequireServiceCredential(next http.HandlerFunc) http.HandlerFunc {
|
||||
return func(w http.ResponseWriter, req *http.Request) {
|
||||
clientID := req.Header.Get("X-Client-Id")
|
||||
secret := req.Header.Get("X-Client-Secret")
|
||||
|
||||
_, ok, err := r.Authenticate(req.Context(), clientID, secret)
|
||||
if err != nil {
|
||||
http.Error(w, "authentifizierung fehlgeschlagen", http.StatusInternalServerError)
|
||||
return
|
||||
}
|
||||
if !ok {
|
||||
http.Error(w, ErrInvalidCredential.Error(), http.StatusUnauthorized)
|
||||
return
|
||||
}
|
||||
next(w, req)
|
||||
}
|
||||
}
|
||||
@@ -1,120 +0,0 @@
|
||||
// Package moduleregistry implementiert Core API-02: die Registry, in der
|
||||
// sich Fachmodule (DMS, Mail, weitere) mit Metadaten eintragen, gekoppelt an
|
||||
// die Aktivierungspruefung aus LIC-02 (Feature-Flags). Zusaetzlich
|
||||
// authentifiziert die Registry Modul-Instanzen selbst ueber ein bei
|
||||
// Provisionierung ausgestelltes Service-Credential.
|
||||
package moduleregistry
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"fmt"
|
||||
|
||||
"github.com/jackc/pgx/v5"
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/flag"
|
||||
)
|
||||
|
||||
var (
|
||||
ErrMissingName = errors.New("moduleregistry: name darf nicht leer sein")
|
||||
ErrMissingVersion = errors.New("moduleregistry: version darf nicht leer sein")
|
||||
ErrModuleNotFound = errors.New("moduleregistry: modul nicht registriert")
|
||||
)
|
||||
|
||||
type Module struct {
|
||||
Name string
|
||||
Version string
|
||||
RequiredFlags []string
|
||||
}
|
||||
|
||||
type Registry struct {
|
||||
pool *pgxpool.Pool
|
||||
flags *flag.Service
|
||||
}
|
||||
|
||||
func NewRegistry(pool *pgxpool.Pool, flags *flag.Service) *Registry {
|
||||
return &Registry{pool: pool, flags: flags}
|
||||
}
|
||||
|
||||
// Register traegt ein Modul mit Name, Version und benoetigten Feature-Flags
|
||||
// ein (Akzeptanzkriterium 1). Fehlende Pflichtangaben werden abgewiesen
|
||||
// (Akzeptanzkriterium 1 / Pruefung 2). Erneutes Register desselben Namens
|
||||
// aktualisiert Version/Flags (Redeploy-Fall).
|
||||
func (r *Registry) Register(ctx context.Context, name, version string, requiredFlags []string) (Module, error) {
|
||||
if name == "" {
|
||||
return Module{}, ErrMissingName
|
||||
}
|
||||
if version == "" {
|
||||
return Module{}, ErrMissingVersion
|
||||
}
|
||||
if requiredFlags == nil {
|
||||
requiredFlags = []string{}
|
||||
}
|
||||
|
||||
_, err := r.pool.Exec(ctx, `
|
||||
INSERT INTO modules (name, version, required_flags, registered_at)
|
||||
VALUES ($1, $2, $3, now())
|
||||
ON CONFLICT (name) DO UPDATE SET version = $2, required_flags = $3, registered_at = now()
|
||||
`, name, version, requiredFlags)
|
||||
if err != nil {
|
||||
return Module{}, fmt.Errorf("modul registrieren: %w", err)
|
||||
}
|
||||
return Module{Name: name, Version: version, RequiredFlags: requiredFlags}, nil
|
||||
}
|
||||
|
||||
func (r *Registry) Get(ctx context.Context, name string) (Module, error) {
|
||||
var m Module
|
||||
m.Name = name
|
||||
err := r.pool.QueryRow(ctx, `
|
||||
SELECT version, required_flags FROM modules WHERE name = $1
|
||||
`, name).Scan(&m.Version, &m.RequiredFlags)
|
||||
if err != nil {
|
||||
if errors.Is(err, pgx.ErrNoRows) {
|
||||
return Module{}, ErrModuleNotFound
|
||||
}
|
||||
return Module{}, fmt.Errorf("modul lesen: %w", err)
|
||||
}
|
||||
return m, nil
|
||||
}
|
||||
|
||||
// List liefert alle registrierten Module (Akzeptanzkriterium 3: ueber API
|
||||
// abfragbar, z.B. fuer Statusseite/Lizenzoberflaeche).
|
||||
func (r *Registry) List(ctx context.Context) ([]Module, error) {
|
||||
rows, err := r.pool.Query(ctx, `SELECT name, version, required_flags FROM modules ORDER BY name`)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("module auflisten: %w", err)
|
||||
}
|
||||
defer rows.Close()
|
||||
|
||||
var out []Module
|
||||
for rows.Next() {
|
||||
var m Module
|
||||
if err := rows.Scan(&m.Name, &m.Version, &m.RequiredFlags); err != nil {
|
||||
return nil, fmt.Errorf("modul lesen: %w", err)
|
||||
}
|
||||
out = append(out, m)
|
||||
}
|
||||
return out, rows.Err()
|
||||
}
|
||||
|
||||
// IsActive prueft, ob ein registriertes Modul fuer einen Tenant aktiviert
|
||||
// ist: registriert UND alle benoetigten Feature-Flags sind fuer diesen
|
||||
// Tenant aktiv (Akzeptanzkriterium 2). Ein nicht registriertes Modul gilt
|
||||
// immer als nicht aktiv.
|
||||
func (r *Registry) IsActive(ctx context.Context, tenantSlug, moduleName string) (bool, error) {
|
||||
m, err := r.Get(ctx, moduleName)
|
||||
if err != nil {
|
||||
if errors.Is(err, ErrModuleNotFound) {
|
||||
return false, nil
|
||||
}
|
||||
return false, err
|
||||
}
|
||||
|
||||
for _, flagKey := range m.RequiredFlags {
|
||||
if !r.flags.IsEnabled(ctx, tenantSlug, flagKey) {
|
||||
return false, nil
|
||||
}
|
||||
}
|
||||
return true, nil
|
||||
}
|
||||
@@ -1,304 +0,0 @@
|
||||
package moduleregistry
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"fmt"
|
||||
"net/http"
|
||||
"net/http/httptest"
|
||||
"os"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
|
||||
"gitea.perlbach24.de/scripte/nexarch/internal/flag"
|
||||
)
|
||||
|
||||
func setupTest(t *testing.T) (*Registry, *flag.Store, func()) {
|
||||
t.Helper()
|
||||
adminDSN := os.Getenv("TEST_ADMIN_DSN")
|
||||
if adminDSN == "" {
|
||||
t.Skip("TEST_ADMIN_DSN nicht gesetzt, Integrationstest uebersprungen")
|
||||
}
|
||||
ctx := context.Background()
|
||||
|
||||
pool, err := pgxpool.New(ctx, adminDSN)
|
||||
if err != nil {
|
||||
t.Fatalf("pool: %v", err)
|
||||
}
|
||||
if _, err := pool.Exec(ctx, `
|
||||
CREATE TABLE IF NOT EXISTS feature_flags (
|
||||
key TEXT PRIMARY KEY, enabled BOOLEAN NOT NULL DEFAULT false,
|
||||
rollout_percentage INT NOT NULL DEFAULT 0, target_tenant_slugs TEXT[] NOT NULL DEFAULT '{}',
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
CREATE TABLE IF NOT EXISTS modules (
|
||||
name TEXT PRIMARY KEY, version TEXT NOT NULL CHECK (version <> ''),
|
||||
required_flags TEXT[] NOT NULL DEFAULT '{}', registered_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
CREATE TABLE IF NOT EXISTS module_credentials (
|
||||
module_name TEXT PRIMARY KEY REFERENCES modules(name),
|
||||
client_id TEXT NOT NULL UNIQUE, secret_hash BYTEA NOT NULL,
|
||||
issued_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
`); err != nil {
|
||||
t.Fatalf("schema: %v", err)
|
||||
}
|
||||
|
||||
flagStore := flag.NewStore(pool)
|
||||
// Kurze TTL, damit Tests, die den Flag-Store direkt aendern (an
|
||||
// Registry.IsActive vorbei), den neuen Stand ohne manuelles Invalidate
|
||||
// zuverlaessig sehen.
|
||||
flagService := flag.NewService(flagStore, 10*time.Millisecond)
|
||||
registry := NewRegistry(pool, flagService)
|
||||
|
||||
cleanup := func() { pool.Close() }
|
||||
return registry, flagStore, cleanup
|
||||
}
|
||||
|
||||
func uniqueModuleName(t *testing.T) string {
|
||||
return fmt.Sprintf("dms_%d", time.Now().UnixNano())
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 1 + Pruefung 2: fehlende Pflichtangaben abgewiesen.
|
||||
func TestRegister_RejectsMissingFields(t *testing.T) {
|
||||
registry, _, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
if _, err := registry.Register(ctx, "", "1.0", nil); !errors.Is(err, ErrMissingName) {
|
||||
t.Fatalf("erwartet ErrMissingName, habe %v", err)
|
||||
}
|
||||
if _, err := registry.Register(ctx, "dms", "", nil); !errors.Is(err, ErrMissingVersion) {
|
||||
t.Fatalf("erwartet ErrMissingVersion, habe %v", err)
|
||||
}
|
||||
}
|
||||
|
||||
func TestRegister_AndGet(t *testing.T) {
|
||||
registry, _, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
name := uniqueModuleName(t)
|
||||
|
||||
m, err := registry.Register(ctx, name, "1.2.0", []string{"dms_enabled"})
|
||||
if err != nil {
|
||||
t.Fatalf("register: %v", err)
|
||||
}
|
||||
if m.Version != "1.2.0" || len(m.RequiredFlags) != 1 {
|
||||
t.Fatalf("unerwartet: %+v", m)
|
||||
}
|
||||
|
||||
got, err := registry.Get(ctx, name)
|
||||
if err != nil {
|
||||
t.Fatalf("get: %v", err)
|
||||
}
|
||||
if got.Version != "1.2.0" {
|
||||
t.Fatalf("get version = %q", got.Version)
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 + 3 + Pruefung 3: konsistente Daten nach
|
||||
// Aktivierung/Deaktivierung eines Moduls.
|
||||
func TestIsActive_ReflectsFlagStateConsistently(t *testing.T) {
|
||||
registry, flagStore, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
name := uniqueModuleName(t)
|
||||
flagKey := name + "_enabled"
|
||||
|
||||
if _, err := registry.Register(ctx, name, "1.0", []string{flagKey}); err != nil {
|
||||
t.Fatalf("register: %v", err)
|
||||
}
|
||||
|
||||
active, err := registry.IsActive(ctx, "acme", name)
|
||||
if err != nil {
|
||||
t.Fatalf("is active (vor flag): %v", err)
|
||||
}
|
||||
if active {
|
||||
t.Fatal("erwartet nicht aktiv, solange flag nicht gesetzt ist")
|
||||
}
|
||||
|
||||
if err := flagStore.Set(ctx, flag.Flag{Key: flagKey, Enabled: true}); err != nil {
|
||||
t.Fatalf("flag setzen: %v", err)
|
||||
}
|
||||
time.Sleep(20 * time.Millisecond) // TTL abwarten
|
||||
|
||||
active, err = registry.IsActive(ctx, "acme", name)
|
||||
if err != nil {
|
||||
t.Fatalf("is active (nach flag an): %v", err)
|
||||
}
|
||||
if !active {
|
||||
t.Fatal("erwartet aktiv, nachdem flag aktiviert wurde")
|
||||
}
|
||||
|
||||
if err := flagStore.Set(ctx, flag.Flag{Key: flagKey, Enabled: false}); err != nil {
|
||||
t.Fatalf("flag zuruecksetzen: %v", err)
|
||||
}
|
||||
time.Sleep(20 * time.Millisecond) // TTL abwarten
|
||||
active, err = registry.IsActive(ctx, "acme", name)
|
||||
if err != nil {
|
||||
t.Fatalf("is active (nach flag aus): %v", err)
|
||||
}
|
||||
if active {
|
||||
t.Fatal("erwartet wieder nicht aktiv, nachdem flag deaktiviert wurde")
|
||||
}
|
||||
}
|
||||
|
||||
func TestIsActive_UnregisteredModuleIsNeverActive(t *testing.T) {
|
||||
registry, _, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
active, err := registry.IsActive(ctx, "acme", "nie-registriert")
|
||||
if err != nil {
|
||||
t.Fatalf("is active: %v", err)
|
||||
}
|
||||
if active {
|
||||
t.Fatal("unregistriertes modul darf nie aktiv sein")
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 + Pruefung 1: Anfrage an deaktiviertes Modul wird
|
||||
// zentral abgewiesen, BEVOR die Modul-Logik erreicht wird.
|
||||
func TestRequireActiveModule_BlocksBeforeHandler(t *testing.T) {
|
||||
registry, flagStore, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
name := uniqueModuleName(t)
|
||||
flagKey := name + "_enabled"
|
||||
|
||||
if _, err := registry.Register(ctx, name, "1.0", []string{flagKey}); err != nil {
|
||||
t.Fatalf("register: %v", err)
|
||||
}
|
||||
|
||||
handlerReached := false
|
||||
handler := registry.RequireActiveModule(name, func(w http.ResponseWriter, r *http.Request) {
|
||||
handlerReached = true
|
||||
w.WriteHeader(http.StatusOK)
|
||||
})
|
||||
|
||||
req := httptest.NewRequest(http.MethodGet, "/modul?tenant=acme", nil)
|
||||
rec := httptest.NewRecorder()
|
||||
handler(rec, req)
|
||||
if rec.Code != http.StatusForbidden {
|
||||
t.Fatalf("status = %d, want 403", rec.Code)
|
||||
}
|
||||
if handlerReached {
|
||||
t.Fatal("handler haette bei deaktiviertem modul NICHT erreicht werden duerfen")
|
||||
}
|
||||
|
||||
if err := flagStore.Set(ctx, flag.Flag{Key: flagKey, Enabled: true}); err != nil {
|
||||
t.Fatalf("flag setzen: %v", err)
|
||||
}
|
||||
req2 := httptest.NewRequest(http.MethodGet, "/modul?tenant=acme", nil)
|
||||
rec2 := httptest.NewRecorder()
|
||||
handler(rec2, req2)
|
||||
if rec2.Code != http.StatusOK {
|
||||
t.Fatalf("status nach aktivierung = %d, want 200", rec2.Code)
|
||||
}
|
||||
if !handlerReached {
|
||||
t.Fatal("handler haette bei aktiviertem modul erreicht werden muessen")
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 4 + Pruefung 4: gueltiges/ungueltiges Service-Credential.
|
||||
func TestProvisionAndAuthenticate(t *testing.T) {
|
||||
registry, _, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
name := uniqueModuleName(t)
|
||||
|
||||
if _, err := registry.Register(ctx, name, "1.0", nil); err != nil {
|
||||
t.Fatalf("register: %v", err)
|
||||
}
|
||||
|
||||
clientID, secret, err := registry.Provision(ctx, name)
|
||||
if err != nil {
|
||||
t.Fatalf("provision: %v", err)
|
||||
}
|
||||
if clientID == "" || secret == "" {
|
||||
t.Fatal("erwartet nicht-leere client-id/secret")
|
||||
}
|
||||
|
||||
moduleName, ok, err := registry.Authenticate(ctx, clientID, secret)
|
||||
if err != nil {
|
||||
t.Fatalf("authenticate (korrekt): %v", err)
|
||||
}
|
||||
if !ok || moduleName != name {
|
||||
t.Fatalf("erwartet erfolgreiche authentifizierung fuer %q, habe ok=%v moduleName=%q", name, ok, moduleName)
|
||||
}
|
||||
|
||||
_, ok, err = registry.Authenticate(ctx, clientID, "falsches-secret")
|
||||
if err != nil {
|
||||
t.Fatalf("authenticate (falsch): %v", err)
|
||||
}
|
||||
if ok {
|
||||
t.Fatal("erwartet fehlschlag bei falschem secret")
|
||||
}
|
||||
|
||||
_, ok, err = registry.Authenticate(ctx, "unbekannte-client-id", secret)
|
||||
if err != nil {
|
||||
t.Fatalf("authenticate (unbekannt): %v", err)
|
||||
}
|
||||
if ok {
|
||||
t.Fatal("erwartet fehlschlag bei unbekannter client-id")
|
||||
}
|
||||
}
|
||||
|
||||
func TestProvision_RequiresRegisteredModule(t *testing.T) {
|
||||
registry, _, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
if _, _, err := registry.Provision(ctx, "nie-registriert"); !errors.Is(err, ErrModuleNotRegistered) {
|
||||
t.Fatalf("erwartet ErrModuleNotRegistered, habe %v", err)
|
||||
}
|
||||
}
|
||||
|
||||
func TestRequireServiceCredential_RejectsInvalidAcceptsValid(t *testing.T) {
|
||||
registry, _, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
name := uniqueModuleName(t)
|
||||
|
||||
if _, err := registry.Register(ctx, name, "1.0", nil); err != nil {
|
||||
t.Fatalf("register: %v", err)
|
||||
}
|
||||
clientID, secret, err := registry.Provision(ctx, name)
|
||||
if err != nil {
|
||||
t.Fatalf("provision: %v", err)
|
||||
}
|
||||
|
||||
handler := registry.RequireServiceCredential(func(w http.ResponseWriter, r *http.Request) {
|
||||
w.WriteHeader(http.StatusOK)
|
||||
})
|
||||
|
||||
// Fehlendes Credential.
|
||||
req := httptest.NewRequest(http.MethodPost, "/service-aufruf", nil)
|
||||
rec := httptest.NewRecorder()
|
||||
handler(rec, req)
|
||||
if rec.Code != http.StatusUnauthorized {
|
||||
t.Fatalf("ohne credential: status = %d, want 401", rec.Code)
|
||||
}
|
||||
|
||||
// Falsches Secret.
|
||||
req2 := httptest.NewRequest(http.MethodPost, "/service-aufruf", nil)
|
||||
req2.Header.Set("X-Client-Id", clientID)
|
||||
req2.Header.Set("X-Client-Secret", "falsch")
|
||||
rec2 := httptest.NewRecorder()
|
||||
handler(rec2, req2)
|
||||
if rec2.Code != http.StatusUnauthorized {
|
||||
t.Fatalf("falsches secret: status = %d, want 401", rec2.Code)
|
||||
}
|
||||
|
||||
// Gueltiges Credential.
|
||||
req3 := httptest.NewRequest(http.MethodPost, "/service-aufruf", nil)
|
||||
req3.Header.Set("X-Client-Id", clientID)
|
||||
req3.Header.Set("X-Client-Secret", secret)
|
||||
rec3 := httptest.NewRecorder()
|
||||
handler(rec3, req3)
|
||||
if rec3.Code != http.StatusOK {
|
||||
t.Fatalf("gueltiges credential: status = %d, want 200", rec3.Code)
|
||||
}
|
||||
}
|
||||
@@ -1,81 +0,0 @@
|
||||
// Package notify implementiert Core CFG-02: den zentralen Benachrichtigungs-
|
||||
// Dispatcher, ueber den beliebige Module Benachrichtigungen ausloesen —
|
||||
// Warteschlange, Wiederholungslogik, Kanal-Abstraktion. Die tatsaechlichen
|
||||
// Kanaele (E-Mail/In-App) sind CFG-03, hier gibt es nur die Sender-
|
||||
// Schnittstelle als Vorbereitung.
|
||||
package notify
|
||||
|
||||
import (
|
||||
"context"
|
||||
"encoding/json"
|
||||
"fmt"
|
||||
"time"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
// DefaultMaxAttempts begrenzt Wiederholungsversuche (Akzeptanzkriterium 2) —
|
||||
// nach dieser Anzahl gibt der Dispatcher kontrolliert auf (status=failed)
|
||||
// statt endlos zu wiederholen.
|
||||
const DefaultMaxAttempts = 5
|
||||
|
||||
// DefaultRetryBackoff ist die Basis-Wartezeit zwischen Wiederholungen,
|
||||
// linear mit der Versuchsnummer skaliert.
|
||||
const DefaultRetryBackoff = 200 * time.Millisecond
|
||||
|
||||
type Notification struct {
|
||||
ID string
|
||||
Channel string
|
||||
Recipient string
|
||||
Payload map[string]any
|
||||
Attempts int
|
||||
}
|
||||
|
||||
// Sender ist die schmale Schnittstelle, die ein konkreter Kanal (CFG-03)
|
||||
// implementiert. Der Dispatcher selbst weiss nichts ueber E-Mail/In-App.
|
||||
type Sender interface {
|
||||
Send(ctx context.Context, n Notification) error
|
||||
}
|
||||
|
||||
// Dispatcher ist die EINE Schnittstelle, ueber die Module Benachrichtigungen
|
||||
// ausloesen — kein Modul baut eigenen Versandcode (Akzeptanzkriterium 1).
|
||||
type Dispatcher struct {
|
||||
pool *pgxpool.Pool
|
||||
maxAttempts int
|
||||
retryBackoff time.Duration
|
||||
}
|
||||
|
||||
func NewDispatcher(pool *pgxpool.Pool) *Dispatcher {
|
||||
return &Dispatcher{pool: pool, maxAttempts: DefaultMaxAttempts, retryBackoff: DefaultRetryBackoff}
|
||||
}
|
||||
|
||||
// WithRetryPolicy erlaubt Tests/Betrieb, Versuchsanzahl und Backoff
|
||||
// anzupassen, ohne die Default-Policy im Produktionscode zu veraendern.
|
||||
func (d *Dispatcher) WithRetryPolicy(maxAttempts int, backoff time.Duration) *Dispatcher {
|
||||
return &Dispatcher{pool: d.pool, maxAttempts: maxAttempts, retryBackoff: backoff}
|
||||
}
|
||||
|
||||
// Enqueue reiht eine Benachrichtigung in die Postgres-Warteschlange ein und
|
||||
// kehrt sofort zurueck — die Zeile ueberlebt jeden Neustart des Dispatcher-
|
||||
// Prozesses unveraendert (Akzeptanzkriterium 3), da sie ausschliesslich in
|
||||
// der Datenbank existiert, nicht im Prozessspeicher.
|
||||
func (d *Dispatcher) Enqueue(ctx context.Context, channel, recipient string, payload map[string]any) (string, error) {
|
||||
if payload == nil {
|
||||
payload = map[string]any{}
|
||||
}
|
||||
payloadJSON, err := json.Marshal(payload)
|
||||
if err != nil {
|
||||
return "", fmt.Errorf("payload serialisieren: %w", err)
|
||||
}
|
||||
|
||||
var id string
|
||||
err = d.pool.QueryRow(ctx, `
|
||||
INSERT INTO notification_jobs (channel, recipient, payload, max_attempts)
|
||||
VALUES ($1, $2, $3, $4)
|
||||
RETURNING id
|
||||
`, channel, recipient, payloadJSON, d.maxAttempts).Scan(&id)
|
||||
if err != nil {
|
||||
return "", fmt.Errorf("benachrichtigung einreihen: %w", err)
|
||||
}
|
||||
return id, nil
|
||||
}
|
||||
@@ -1,219 +0,0 @@
|
||||
package notify
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"os"
|
||||
"sync"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
func setupTest(t *testing.T) (*pgxpool.Pool, func()) {
|
||||
t.Helper()
|
||||
adminDSN := os.Getenv("TEST_ADMIN_DSN")
|
||||
if adminDSN == "" {
|
||||
t.Skip("TEST_ADMIN_DSN nicht gesetzt, Integrationstest uebersprungen")
|
||||
}
|
||||
ctx := context.Background()
|
||||
|
||||
pool, err := pgxpool.New(ctx, adminDSN)
|
||||
if err != nil {
|
||||
t.Fatalf("pool: %v", err)
|
||||
}
|
||||
if _, err := pool.Exec(ctx, `
|
||||
CREATE TABLE IF NOT EXISTS notification_jobs (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
channel TEXT NOT NULL,
|
||||
recipient TEXT NOT NULL,
|
||||
payload JSONB NOT NULL DEFAULT '{}'::jsonb,
|
||||
status TEXT NOT NULL DEFAULT 'pending' CHECK (status IN ('pending', 'sent', 'failed')),
|
||||
attempts INT NOT NULL DEFAULT 0,
|
||||
max_attempts INT NOT NULL DEFAULT 5,
|
||||
next_attempt_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
last_error TEXT,
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
)`); err != nil {
|
||||
t.Fatalf("schema: %v", err)
|
||||
}
|
||||
|
||||
cleanup := func() { pool.Close() }
|
||||
return pool, cleanup
|
||||
}
|
||||
|
||||
type fakeSender struct {
|
||||
mu sync.Mutex
|
||||
sentIDs []string
|
||||
failUntil int
|
||||
calls int
|
||||
}
|
||||
|
||||
func (f *fakeSender) Send(ctx context.Context, n Notification) error {
|
||||
f.mu.Lock()
|
||||
defer f.mu.Unlock()
|
||||
f.calls++
|
||||
if f.calls <= f.failUntil {
|
||||
return errors.New("simulierter zustellfehler")
|
||||
}
|
||||
f.sentIDs = append(f.sentIDs, n.ID)
|
||||
return nil
|
||||
}
|
||||
|
||||
func (f *fakeSender) sentCount() int {
|
||||
f.mu.Lock()
|
||||
defer f.mu.Unlock()
|
||||
return len(f.sentIDs)
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 1: Module loesen ueber Enqueue aus, keine eigene
|
||||
// Versandlogik noetig.
|
||||
func TestDispatcher_EnqueueAndProcess(t *testing.T) {
|
||||
pool, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
d := NewDispatcher(pool)
|
||||
id, err := d.Enqueue(ctx, "email", "alice@example.com", map[string]any{"subject": "Willkommen"})
|
||||
if err != nil {
|
||||
t.Fatalf("enqueue: %v", err)
|
||||
}
|
||||
if id == "" {
|
||||
t.Fatal("erwartet nicht-leere id")
|
||||
}
|
||||
|
||||
sender := &fakeSender{}
|
||||
sent, failed, err := d.ProcessDue(ctx, sender, 10)
|
||||
if err != nil {
|
||||
t.Fatalf("process: %v", err)
|
||||
}
|
||||
if sent != 1 || failed != 0 {
|
||||
t.Fatalf("erwartet sent=1 failed=0, habe sent=%d failed=%d", sent, failed)
|
||||
}
|
||||
if sender.sentCount() != 1 {
|
||||
t.Fatalf("erwartet 1 zustellung, habe %d", sender.sentCount())
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 + Pruefung 2: Wiederholungslogik greift bei
|
||||
// simuliertem Fehler und bricht nach definierter Anzahl kontrolliert ab.
|
||||
func TestProcessDue_RetriesThenGivesUpAfterMaxAttempts(t *testing.T) {
|
||||
pool, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
d := NewDispatcher(pool).WithRetryPolicy(3, time.Millisecond)
|
||||
id, err := d.Enqueue(ctx, "email", "bob@example.com", nil)
|
||||
if err != nil {
|
||||
t.Fatalf("enqueue: %v", err)
|
||||
}
|
||||
|
||||
sender := &fakeSender{failUntil: 100} // schlaegt bei jedem versuch fehl
|
||||
|
||||
for i := 0; i < 3; i++ {
|
||||
time.Sleep(5 * time.Millisecond) // next_attempt_at abwarten
|
||||
if _, _, err := d.ProcessDue(ctx, sender, 10); err != nil {
|
||||
t.Fatalf("process %d: %v", i, err)
|
||||
}
|
||||
}
|
||||
|
||||
var status string
|
||||
var attempts int
|
||||
if err := pool.QueryRow(ctx, `SELECT status, attempts FROM notification_jobs WHERE id = $1`, id).Scan(&status, &attempts); err != nil {
|
||||
t.Fatalf("status lesen: %v", err)
|
||||
}
|
||||
if status != "failed" {
|
||||
t.Fatalf("erwartet status failed nach max_attempts, habe %q", status)
|
||||
}
|
||||
if attempts != 3 {
|
||||
t.Fatalf("erwartet 3 versuche, habe %d", attempts)
|
||||
}
|
||||
|
||||
// Weiteres ProcessDue darf den bereits aufgegebenen job nicht mehr anfassen.
|
||||
sent, failed, err := d.ProcessDue(ctx, sender, 10)
|
||||
if err != nil {
|
||||
t.Fatalf("process nach abbruch: %v", err)
|
||||
}
|
||||
if sent != 0 || failed != 0 {
|
||||
t.Fatalf("erwartet keine weitere verarbeitung, habe sent=%d failed=%d", sent, failed)
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 3 + Pruefung 1: Neustart des Dienstes waehrend offener
|
||||
// Zustellung verliert keine Nachricht — simuliert durch eine komplett neue
|
||||
// Dispatcher/Pool-Instanz nach dem Enqueue, bevor irgendetwas verarbeitet wurde.
|
||||
func TestQueue_SurvivesRestartWithoutMessageLoss(t *testing.T) {
|
||||
pool, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
firstInstance := NewDispatcher(pool)
|
||||
id, err := firstInstance.Enqueue(ctx, "email", "carol@example.com", nil)
|
||||
if err != nil {
|
||||
t.Fatalf("enqueue: %v", err)
|
||||
}
|
||||
|
||||
// "Neustart": eine voellig neue Dispatcher-Instanz (repraesentiert einen
|
||||
// neuen Prozess) verbindet sich neu und verarbeitet die Warteschlange —
|
||||
// die Nachricht existiert ausschliesslich in Postgres, nicht im
|
||||
// Prozessspeicher der ersten Instanz.
|
||||
restartedInstance := NewDispatcher(pool)
|
||||
sender := &fakeSender{}
|
||||
sent, failed, err := restartedInstance.ProcessDue(ctx, sender, 10)
|
||||
if err != nil {
|
||||
t.Fatalf("process nach neustart: %v", err)
|
||||
}
|
||||
if sent != 1 || failed != 0 {
|
||||
t.Fatalf("erwartet sent=1 nach neustart, habe sent=%d failed=%d", sent, failed)
|
||||
}
|
||||
if len(sender.sentIDs) != 1 || sender.sentIDs[0] != id {
|
||||
t.Fatalf("erwartet zustellung der urspruenglichen nachricht %q, habe %v", id, sender.sentIDs)
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 3 + Pruefung 3: zwei gleichzeitig ausloesende Module,
|
||||
// beide Nachrichten werden korrekt (und nicht doppelt) zugestellt.
|
||||
func TestProcessDue_ConcurrentDispatchBothDelivered(t *testing.T) {
|
||||
pool, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
d := NewDispatcher(pool)
|
||||
idA, err := d.Enqueue(ctx, "email", "modul-a@example.com", nil)
|
||||
if err != nil {
|
||||
t.Fatalf("enqueue a: %v", err)
|
||||
}
|
||||
idB, err := d.Enqueue(ctx, "email", "modul-b@example.com", nil)
|
||||
if err != nil {
|
||||
t.Fatalf("enqueue b: %v", err)
|
||||
}
|
||||
|
||||
sender := &fakeSender{}
|
||||
var wg sync.WaitGroup
|
||||
for i := 0; i < 2; i++ {
|
||||
wg.Add(1)
|
||||
go func() {
|
||||
defer wg.Done()
|
||||
if _, _, err := d.ProcessDue(ctx, sender, 10); err != nil {
|
||||
t.Errorf("process: %v", err)
|
||||
}
|
||||
}()
|
||||
}
|
||||
wg.Wait()
|
||||
|
||||
if sender.sentCount() != 2 {
|
||||
t.Fatalf("erwartet genau 2 zustellungen, habe %d: %v", sender.sentCount(), sender.sentIDs)
|
||||
}
|
||||
seen := map[string]bool{}
|
||||
for _, id := range sender.sentIDs {
|
||||
if seen[id] {
|
||||
t.Fatalf("nachricht %q wurde doppelt zugestellt", id)
|
||||
}
|
||||
seen[id] = true
|
||||
}
|
||||
if !seen[idA] || !seen[idB] {
|
||||
t.Fatalf("erwartet beide nachrichten zugestellt, habe %v", sender.sentIDs)
|
||||
}
|
||||
}
|
||||
@@ -1,102 +0,0 @@
|
||||
package notify
|
||||
|
||||
import (
|
||||
"context"
|
||||
"encoding/json"
|
||||
"fmt"
|
||||
"time"
|
||||
)
|
||||
|
||||
// ProcessDue holt bis zu limit faellige Benachrichtigungen und versucht sie
|
||||
// ueber sender zuzustellen. FOR UPDATE SKIP LOCKED serialisiert konkurrierende
|
||||
// Aufrufe (Akzeptanzkriterium 3 / Pruefung 3: zwei gleichzeitig ausloesende
|
||||
// Module duerfen sich nicht gegenseitig blockieren oder Nachrichten doppelt
|
||||
// zustellen) — dieselbe Konvention wie internal/tenant.Lifecycle.ProcessDueDeletions.
|
||||
func (d *Dispatcher) ProcessDue(ctx context.Context, sender Sender, limit int) (sent, failed int, err error) {
|
||||
tx, err := d.pool.Begin(ctx)
|
||||
if err != nil {
|
||||
return 0, 0, fmt.Errorf("transaktion starten: %w", err)
|
||||
}
|
||||
defer func() { _ = tx.Rollback(ctx) }()
|
||||
|
||||
rows, err := tx.Query(ctx, `
|
||||
SELECT id, channel, recipient, payload, attempts, max_attempts
|
||||
FROM notification_jobs
|
||||
WHERE status = 'pending' AND next_attempt_at <= now()
|
||||
ORDER BY created_at
|
||||
FOR UPDATE SKIP LOCKED
|
||||
LIMIT $1
|
||||
`, limit)
|
||||
if err != nil {
|
||||
return 0, 0, fmt.Errorf("faellige benachrichtigungen abfragen: %w", err)
|
||||
}
|
||||
|
||||
type due struct {
|
||||
id, channel, recipient string
|
||||
payload []byte
|
||||
attempts, maxAttempts int
|
||||
}
|
||||
var candidates []due
|
||||
for rows.Next() {
|
||||
var c due
|
||||
if err := rows.Scan(&c.id, &c.channel, &c.recipient, &c.payload, &c.attempts, &c.maxAttempts); err != nil {
|
||||
rows.Close()
|
||||
return 0, 0, fmt.Errorf("faellige benachrichtigung lesen: %w", err)
|
||||
}
|
||||
candidates = append(candidates, c)
|
||||
}
|
||||
rows.Close()
|
||||
if err := rows.Err(); err != nil {
|
||||
return 0, 0, err
|
||||
}
|
||||
|
||||
for _, c := range candidates {
|
||||
var payload map[string]any
|
||||
if err := json.Unmarshal(c.payload, &payload); err != nil {
|
||||
payload = map[string]any{}
|
||||
}
|
||||
|
||||
sendErr := sender.Send(ctx, Notification{
|
||||
ID: c.id, Channel: c.channel, Recipient: c.recipient, Payload: payload, Attempts: c.attempts,
|
||||
})
|
||||
|
||||
if sendErr == nil {
|
||||
if _, err := tx.Exec(ctx, `
|
||||
UPDATE notification_jobs SET status = 'sent', updated_at = now() WHERE id = $1
|
||||
`, c.id); err != nil {
|
||||
return sent, failed, fmt.Errorf("erfolg speichern: %w", err)
|
||||
}
|
||||
sent++
|
||||
continue
|
||||
}
|
||||
|
||||
newAttempts := c.attempts + 1
|
||||
if newAttempts >= c.maxAttempts {
|
||||
// Akzeptanzkriterium 2: kontrollierter Abbruch nach definierter
|
||||
// Anzahl Versuche, kein endloses Wiederholen.
|
||||
if _, err := tx.Exec(ctx, `
|
||||
UPDATE notification_jobs
|
||||
SET status = 'failed', attempts = $2, last_error = $3, updated_at = now()
|
||||
WHERE id = $1
|
||||
`, c.id, newAttempts, sendErr.Error()); err != nil {
|
||||
return sent, failed, fmt.Errorf("fehlschlag speichern: %w", err)
|
||||
}
|
||||
failed++
|
||||
continue
|
||||
}
|
||||
|
||||
nextAttempt := time.Now().Add(time.Duration(newAttempts) * d.retryBackoff)
|
||||
if _, err := tx.Exec(ctx, `
|
||||
UPDATE notification_jobs
|
||||
SET attempts = $2, next_attempt_at = $3, last_error = $4, updated_at = now()
|
||||
WHERE id = $1
|
||||
`, c.id, newAttempts, nextAttempt, sendErr.Error()); err != nil {
|
||||
return sent, failed, fmt.Errorf("wiederholung planen: %w", err)
|
||||
}
|
||||
}
|
||||
|
||||
if err := tx.Commit(ctx); err != nil {
|
||||
return 0, 0, fmt.Errorf("transaktion committen: %w", err)
|
||||
}
|
||||
return sent, failed, nil
|
||||
}
|
||||
@@ -0,0 +1,66 @@
|
||||
package user
|
||||
|
||||
import (
|
||||
"encoding/json"
|
||||
"errors"
|
||||
"net/http"
|
||||
)
|
||||
|
||||
// Handler stellt die CRUD-API fuer Benutzerkonten bereit (IAM-01-Auftrag).
|
||||
// Auth/Sessions (IAM-02) und Rollen (RBAC-01) sind ausdruecklich nicht Teil
|
||||
// dieser Kachel und daher hier noch nicht angebunden.
|
||||
type Handler struct {
|
||||
users *TenantUserStore
|
||||
superadmins *SuperadminStore
|
||||
}
|
||||
|
||||
func NewHandler(users *TenantUserStore, superadmins *SuperadminStore) *Handler {
|
||||
return &Handler{users: users, superadmins: superadmins}
|
||||
}
|
||||
|
||||
type createUserRequest struct {
|
||||
Email string `json:"email"`
|
||||
Name string `json:"name"`
|
||||
}
|
||||
|
||||
func (h *Handler) CreateUser(w http.ResponseWriter, r *http.Request) {
|
||||
var req createUserRequest
|
||||
if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
|
||||
http.Error(w, "ungueltige Anfrage", http.StatusBadRequest)
|
||||
return
|
||||
}
|
||||
|
||||
u, err := h.users.Create(r.Context(), req.Email, req.Name)
|
||||
writeUserResult(w, u, err)
|
||||
}
|
||||
|
||||
// CreateSuperadmin legt ein mandantenuebergreifendes Superadmin-Konto an —
|
||||
// bewusst ein eigener Endpunkt statt eines Tenant-Parameters mit Null-Wert.
|
||||
func (h *Handler) CreateSuperadmin(w http.ResponseWriter, r *http.Request) {
|
||||
var req createUserRequest
|
||||
if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
|
||||
http.Error(w, "ungueltige Anfrage", http.StatusBadRequest)
|
||||
return
|
||||
}
|
||||
|
||||
u, err := h.superadmins.Create(r.Context(), req.Email, req.Name)
|
||||
writeUserResult(w, u, err)
|
||||
}
|
||||
|
||||
func writeUserResult(w http.ResponseWriter, u User, err error) {
|
||||
if err != nil {
|
||||
switch {
|
||||
case errors.Is(err, ErrInvalidEmail), errors.Is(err, ErrEmailTaken):
|
||||
http.Error(w, err.Error(), http.StatusBadRequest)
|
||||
case errors.Is(err, ErrNotFound):
|
||||
http.Error(w, err.Error(), http.StatusNotFound)
|
||||
default:
|
||||
http.Error(w, "benutzer konnte nicht verarbeitet werden", http.StatusInternalServerError)
|
||||
}
|
||||
return
|
||||
}
|
||||
|
||||
w.Header().Set("Content-Type", "application/json")
|
||||
w.WriteHeader(http.StatusCreated)
|
||||
_ = json.NewEncoder(w).Encode(u)
|
||||
}
|
||||
@@ -0,0 +1,173 @@
|
||||
package user
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"fmt"
|
||||
"os"
|
||||
"strings"
|
||||
"testing"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
// setupTestDB legt eine frische, isolierte Testdatenbank an, wendet die
|
||||
// uebergebene Migration an und liefert einen verbundenen Pool. Wird ohne
|
||||
// TEST_ADMIN_DSN uebersprungen — siehe internal/tenant/provisioner_test.go
|
||||
// fuer dasselbe Muster.
|
||||
func setupTestDB(t *testing.T, dbName, schemaSQL string) *pgxpool.Pool {
|
||||
t.Helper()
|
||||
|
||||
adminDSN := os.Getenv("TEST_ADMIN_DSN")
|
||||
if adminDSN == "" {
|
||||
t.Skip("TEST_ADMIN_DSN nicht gesetzt, Integrationstest uebersprungen")
|
||||
}
|
||||
ctx := context.Background()
|
||||
|
||||
adminPool, err := pgxpool.New(ctx, adminDSN)
|
||||
if err != nil {
|
||||
t.Fatalf("admin pool: %v", err)
|
||||
}
|
||||
|
||||
_, _ = adminPool.Exec(ctx, fmt.Sprintf(`DROP DATABASE IF EXISTS %q`, dbName))
|
||||
if _, err := adminPool.Exec(ctx, fmt.Sprintf(`CREATE DATABASE %q`, dbName)); err != nil {
|
||||
t.Fatalf("testdatenbank anlegen: %v", err)
|
||||
}
|
||||
|
||||
dsn := strings.Replace(adminDSN, "/postgres?", "/"+dbName+"?", 1)
|
||||
pool, err := pgxpool.New(ctx, dsn)
|
||||
if err != nil {
|
||||
t.Fatalf("connect testdatenbank: %v", err)
|
||||
}
|
||||
|
||||
if _, err := pool.Exec(ctx, schemaSQL); err != nil {
|
||||
t.Fatalf("schema anwenden: %v", err)
|
||||
}
|
||||
|
||||
t.Cleanup(func() {
|
||||
pool.Close()
|
||||
_, _ = adminPool.Exec(ctx, fmt.Sprintf(`DROP DATABASE IF EXISTS %q`, dbName))
|
||||
adminPool.Close()
|
||||
})
|
||||
|
||||
return pool
|
||||
}
|
||||
|
||||
const usersSchema = `
|
||||
CREATE EXTENSION IF NOT EXISTS pgcrypto;
|
||||
CREATE TABLE users (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
email TEXT NOT NULL UNIQUE,
|
||||
name TEXT NOT NULL,
|
||||
status TEXT NOT NULL DEFAULT 'active',
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);`
|
||||
|
||||
const superadminsSchema = `
|
||||
CREATE EXTENSION IF NOT EXISTS pgcrypto;
|
||||
CREATE TABLE superadmins (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
email TEXT NOT NULL UNIQUE,
|
||||
name TEXT NOT NULL,
|
||||
status TEXT NOT NULL DEFAULT 'active',
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);`
|
||||
|
||||
// Akzeptanzkriterium 1 + 3, Pruefung 1 (inkl. Negativfall doppelte E-Mail).
|
||||
func TestTenantUserStore_CRUD(t *testing.T) {
|
||||
pool := setupTestDB(t, "test_iam01_users", usersSchema)
|
||||
store := NewTenantUserStore(pool)
|
||||
ctx := context.Background()
|
||||
|
||||
created, err := store.Create(ctx, "alice@example.com", "Alice")
|
||||
if err != nil {
|
||||
t.Fatalf("create: %v", err)
|
||||
}
|
||||
if created.Status != StatusActive {
|
||||
t.Fatalf("erwartet status active, hat %q", created.Status)
|
||||
}
|
||||
|
||||
got, err := store.Get(ctx, created.ID)
|
||||
if err != nil {
|
||||
t.Fatalf("get: %v", err)
|
||||
}
|
||||
if got.Email != "alice@example.com" {
|
||||
t.Fatalf("get email = %q", got.Email)
|
||||
}
|
||||
|
||||
updated, err := store.Update(ctx, created.ID, "", "Alice A.")
|
||||
if err != nil {
|
||||
t.Fatalf("update: %v", err)
|
||||
}
|
||||
if updated.Name != "Alice A." || updated.Email != "alice@example.com" {
|
||||
t.Fatalf("update ergebnis unerwartet: %+v", updated)
|
||||
}
|
||||
|
||||
list, err := store.List(ctx)
|
||||
if err != nil {
|
||||
t.Fatalf("list: %v", err)
|
||||
}
|
||||
if len(list) != 1 {
|
||||
t.Fatalf("erwartet 1 benutzer, habe %d", len(list))
|
||||
}
|
||||
|
||||
deactivated, err := store.Deactivate(ctx, created.ID)
|
||||
if err != nil {
|
||||
t.Fatalf("deactivate: %v", err)
|
||||
}
|
||||
if deactivated.Status != StatusInactive {
|
||||
t.Fatalf("erwartet status inactive, hat %q", deactivated.Status)
|
||||
}
|
||||
|
||||
// Negativfall: doppelte E-Mail-Adresse.
|
||||
if _, err := store.Create(ctx, "second@example.com", "Bob"); err != nil {
|
||||
t.Fatalf("create second: %v", err)
|
||||
}
|
||||
if _, err := store.Create(ctx, "second@example.com", "Bob Zwei"); !errors.Is(err, ErrEmailTaken) {
|
||||
t.Fatalf("erwartet ErrEmailTaken, habe %v", err)
|
||||
}
|
||||
|
||||
// Negativfall: fehlender Benutzer.
|
||||
if _, err := store.Get(ctx, created.ID+"-nicht-vorhanden"); err == nil {
|
||||
t.Fatalf("erwartet fehler bei unbekannter/ungueltiger id")
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 + Pruefung 2: Superadmin-Anlage ohne Tenant-Kontext.
|
||||
// SuperadminStore.Create hat keinen Tenant-Parameter — es gibt syntaktisch
|
||||
// keine Moeglichkeit, hier versehentlich einen Tenant-Sonderfall zu vergessen.
|
||||
func TestSuperadminStore_CreateWithoutTenantContext(t *testing.T) {
|
||||
pool := setupTestDB(t, "test_iam01_superadmins", superadminsSchema)
|
||||
store := NewSuperadminStore(pool)
|
||||
ctx := context.Background()
|
||||
|
||||
created, err := store.Create(ctx, "root@nexarch.internal", "Root")
|
||||
if err != nil {
|
||||
t.Fatalf("create superadmin: %v", err)
|
||||
}
|
||||
if created.Status != StatusActive {
|
||||
t.Fatalf("erwartet status active, hat %q", created.Status)
|
||||
}
|
||||
|
||||
got, err := store.Get(ctx, created.ID)
|
||||
if err != nil {
|
||||
t.Fatalf("get: %v", err)
|
||||
}
|
||||
if got.Email != "root@nexarch.internal" {
|
||||
t.Fatalf("get email = %q", got.Email)
|
||||
}
|
||||
|
||||
if _, err := store.Create(ctx, "root@nexarch.internal", "Root Zwei"); !errors.Is(err, ErrEmailTaken) {
|
||||
t.Fatalf("erwartet ErrEmailTaken (globale eindeutigkeit), habe %v", err)
|
||||
}
|
||||
|
||||
deactivated, err := store.Deactivate(ctx, created.ID)
|
||||
if err != nil {
|
||||
t.Fatalf("deactivate: %v", err)
|
||||
}
|
||||
if deactivated.Status != StatusInactive {
|
||||
t.Fatalf("erwartet status inactive, hat %q", deactivated.Status)
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,77 @@
|
||||
package user
|
||||
|
||||
import (
|
||||
"context"
|
||||
"fmt"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
// SuperadminStore verwaltet mandantenuebergreifende Superadmin-Konten in der
|
||||
// Control-Plane-Registry (siehe internal/tenant.Registry). Superadmin-ohne-
|
||||
// Tenant ist dadurch ein eigener Typ statt eines Sonderfalls von User/
|
||||
// TenantUserStore — es gibt keinen Tenant-Parameter, den man weglassen
|
||||
// koennte (IAM-01, "ohne Sonderbehandlung im Code").
|
||||
type SuperadminStore struct {
|
||||
pool *pgxpool.Pool
|
||||
}
|
||||
|
||||
func NewSuperadminStore(pool *pgxpool.Pool) *SuperadminStore {
|
||||
return &SuperadminStore{pool: pool}
|
||||
}
|
||||
|
||||
func (s *SuperadminStore) Create(ctx context.Context, email, name string) (User, error) {
|
||||
if err := ValidateEmail(email); err != nil {
|
||||
return User{}, err
|
||||
}
|
||||
|
||||
var u User
|
||||
u.Email, u.Name, u.Status = email, name, StatusActive
|
||||
|
||||
row := s.pool.QueryRow(ctx, `
|
||||
INSERT INTO superadmins (email, name, status)
|
||||
VALUES ($1, $2, $3)
|
||||
RETURNING id, created_at, updated_at
|
||||
`, u.Email, u.Name, u.Status)
|
||||
|
||||
if err := row.Scan(&u.ID, &u.CreatedAt, &u.UpdatedAt); err != nil {
|
||||
return User{}, mapWriteErr(err)
|
||||
}
|
||||
return u, nil
|
||||
}
|
||||
|
||||
func (s *SuperadminStore) Get(ctx context.Context, id string) (User, error) {
|
||||
return scanUser(s.pool.QueryRow(ctx, `
|
||||
SELECT id, email, name, status, created_at, updated_at
|
||||
FROM superadmins WHERE id = $1
|
||||
`, id))
|
||||
}
|
||||
|
||||
func (s *SuperadminStore) List(ctx context.Context) ([]User, error) {
|
||||
rows, err := s.pool.Query(ctx, `
|
||||
SELECT id, email, name, status, created_at, updated_at
|
||||
FROM superadmins ORDER BY created_at
|
||||
`)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("superadmins auflisten: %w", err)
|
||||
}
|
||||
defer rows.Close()
|
||||
|
||||
var out []User
|
||||
for rows.Next() {
|
||||
var u User
|
||||
if err := rows.Scan(&u.ID, &u.Email, &u.Name, &u.Status, &u.CreatedAt, &u.UpdatedAt); err != nil {
|
||||
return nil, fmt.Errorf("superadmin lesen: %w", err)
|
||||
}
|
||||
out = append(out, u)
|
||||
}
|
||||
return out, rows.Err()
|
||||
}
|
||||
|
||||
func (s *SuperadminStore) Deactivate(ctx context.Context, id string) (User, error) {
|
||||
return scanUser(s.pool.QueryRow(ctx, `
|
||||
UPDATE superadmins SET status = $2, updated_at = now()
|
||||
WHERE id = $1
|
||||
RETURNING id, email, name, status, created_at, updated_at
|
||||
`, id, StatusInactive))
|
||||
}
|
||||
@@ -0,0 +1,173 @@
|
||||
package user
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"fmt"
|
||||
|
||||
"github.com/jackc/pgx/v5"
|
||||
"github.com/jackc/pgx/v5/pgconn"
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
// TenantUserStore verwaltet Benutzer innerhalb GENAU EINER Tenant-Datenbank.
|
||||
// Welcher Mandant gemeint ist, ergibt sich ausschliesslich aus dem
|
||||
// uebergebenen Pool — es gibt keine tenant_id-Spalte (siehe migrations/tenant/0001_users.up.sql).
|
||||
type TenantUserStore struct {
|
||||
pool *pgxpool.Pool
|
||||
}
|
||||
|
||||
func NewTenantUserStore(pool *pgxpool.Pool) *TenantUserStore {
|
||||
return &TenantUserStore{pool: pool}
|
||||
}
|
||||
|
||||
func (s *TenantUserStore) Create(ctx context.Context, email, name string) (User, error) {
|
||||
if err := ValidateEmail(email); err != nil {
|
||||
return User{}, err
|
||||
}
|
||||
|
||||
var u User
|
||||
u.Email, u.Name, u.Status = email, name, StatusActive
|
||||
|
||||
row := s.pool.QueryRow(ctx, `
|
||||
INSERT INTO users (email, name, status)
|
||||
VALUES ($1, $2, $3)
|
||||
RETURNING id, created_at, updated_at
|
||||
`, u.Email, u.Name, u.Status)
|
||||
|
||||
if err := row.Scan(&u.ID, &u.CreatedAt, &u.UpdatedAt); err != nil {
|
||||
return User{}, mapWriteErr(err)
|
||||
}
|
||||
return u, nil
|
||||
}
|
||||
|
||||
func (s *TenantUserStore) Get(ctx context.Context, id string) (User, error) {
|
||||
return scanUser(s.pool.QueryRow(ctx, `
|
||||
SELECT id, email, name, status, created_at, updated_at
|
||||
FROM users WHERE id = $1
|
||||
`, id))
|
||||
}
|
||||
|
||||
func (s *TenantUserStore) List(ctx context.Context) ([]User, error) {
|
||||
rows, err := s.pool.Query(ctx, `
|
||||
SELECT id, email, name, status, created_at, updated_at
|
||||
FROM users ORDER BY created_at
|
||||
`)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("benutzer auflisten: %w", err)
|
||||
}
|
||||
defer rows.Close()
|
||||
|
||||
var out []User
|
||||
for rows.Next() {
|
||||
var u User
|
||||
if err := rows.Scan(&u.ID, &u.Email, &u.Name, &u.Status, &u.CreatedAt, &u.UpdatedAt); err != nil {
|
||||
return nil, fmt.Errorf("benutzer lesen: %w", err)
|
||||
}
|
||||
out = append(out, u)
|
||||
}
|
||||
return out, rows.Err()
|
||||
}
|
||||
|
||||
// Update aendert Name und E-Mail. Eine leere email/name laesst das jeweilige
|
||||
// Feld unveraendert.
|
||||
func (s *TenantUserStore) Update(ctx context.Context, id, email, name string) (User, error) {
|
||||
if email != "" {
|
||||
if err := ValidateEmail(email); err != nil {
|
||||
return User{}, err
|
||||
}
|
||||
}
|
||||
|
||||
row := s.pool.QueryRow(ctx, `
|
||||
UPDATE users
|
||||
SET email = COALESCE(NULLIF($2, ''), email),
|
||||
name = COALESCE(NULLIF($3, ''), name),
|
||||
updated_at = now()
|
||||
WHERE id = $1
|
||||
RETURNING id, email, name, status, created_at, updated_at
|
||||
`, id, email, name)
|
||||
|
||||
u, err := scanUser(row)
|
||||
if err != nil {
|
||||
return User{}, mapWriteErr(err)
|
||||
}
|
||||
return u, nil
|
||||
}
|
||||
|
||||
// Deactivate setzt den Benutzer auf inaktiv statt ihn zu loeschen.
|
||||
func (s *TenantUserStore) Deactivate(ctx context.Context, id string) (User, error) {
|
||||
return scanUser(s.pool.QueryRow(ctx, `
|
||||
UPDATE users SET status = $2, updated_at = now()
|
||||
WHERE id = $1
|
||||
RETURNING id, email, name, status, created_at, updated_at
|
||||
`, id, StatusInactive))
|
||||
}
|
||||
|
||||
// SetPasswordHash schreibt einen bereits berechneten bcrypt-Hash (siehe
|
||||
// internal/auth, IAM-02). Der Store selbst kennt kein Klartext-Passwort.
|
||||
func (s *TenantUserStore) SetPasswordHash(ctx context.Context, id, hash string) error {
|
||||
tag, err := s.pool.Exec(ctx, `
|
||||
UPDATE users SET password_hash = $2, updated_at = now() WHERE id = $1
|
||||
`, id, hash)
|
||||
if err != nil {
|
||||
return fmt.Errorf("passwort setzen: %w", err)
|
||||
}
|
||||
if tag.RowsAffected() == 0 {
|
||||
return ErrNotFound
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
// AuthCredentials wird ausschliesslich fuer den Login-Pfad (internal/auth)
|
||||
// verwendet und traegt bewusst den password_hash, damit er nicht ueber den
|
||||
// regulaeren User-Typ/JSON-Serialisierungspfad nach aussen dringen kann.
|
||||
type AuthCredentials struct {
|
||||
User User
|
||||
PasswordHash string
|
||||
}
|
||||
|
||||
// GetByEmailForAuth liefert Benutzer + Passwort-Hash zu einer E-Mail-Adresse
|
||||
// aus GENAU DIESER Tenant-Datenbank — der Tenant-Scope ergibt sich damit
|
||||
// zwingend aus dem verwendeten Pool, es gibt keine Moeglichkeit, versehentlich
|
||||
// ueber Tenant-Grenzen hinweg zu suchen (bekannter archivmail-Fehler, siehe
|
||||
// IAM-02 "Bekannte Fehler vermeiden").
|
||||
func (s *TenantUserStore) GetByEmailForAuth(ctx context.Context, email string) (AuthCredentials, error) {
|
||||
var c AuthCredentials
|
||||
row := s.pool.QueryRow(ctx, `
|
||||
SELECT id, email, name, status, created_at, updated_at, password_hash
|
||||
FROM users WHERE email = $1
|
||||
`, email)
|
||||
|
||||
if err := row.Scan(&c.User.ID, &c.User.Email, &c.User.Name, &c.User.Status,
|
||||
&c.User.CreatedAt, &c.User.UpdatedAt, &c.PasswordHash); err != nil {
|
||||
if errors.Is(err, pgx.ErrNoRows) {
|
||||
return AuthCredentials{}, ErrNotFound
|
||||
}
|
||||
return AuthCredentials{}, fmt.Errorf("anmeldedaten lesen: %w", err)
|
||||
}
|
||||
return c, nil
|
||||
}
|
||||
|
||||
func scanUser(row pgx.Row) (User, error) {
|
||||
var u User
|
||||
if err := row.Scan(&u.ID, &u.Email, &u.Name, &u.Status, &u.CreatedAt, &u.UpdatedAt); err != nil {
|
||||
if errors.Is(err, pgx.ErrNoRows) {
|
||||
return User{}, ErrNotFound
|
||||
}
|
||||
return User{}, fmt.Errorf("benutzer lesen: %w", err)
|
||||
}
|
||||
return u, nil
|
||||
}
|
||||
|
||||
// mapWriteErr uebersetzt den Unique-Constraint-Verstoss der E-Mail-Spalte in
|
||||
// einen sprechenden Fehler statt der rohen Postgres-Fehlermeldung.
|
||||
func mapWriteErr(err error) error {
|
||||
var pgErr *pgconn.PgError
|
||||
if errors.As(err, &pgErr) && pgErr.Code == "23505" {
|
||||
return ErrEmailTaken
|
||||
}
|
||||
if errors.Is(err, pgx.ErrNoRows) {
|
||||
return ErrNotFound
|
||||
}
|
||||
return fmt.Errorf("benutzer schreiben: %w", err)
|
||||
}
|
||||
@@ -0,0 +1,42 @@
|
||||
// Package user implementiert Core IAM-01: das Benutzer-Datenmodell und die
|
||||
// CRUD-Operationen. Tenant-Zugehoerigkeit ist ueber die Zieldatenbank
|
||||
// gegeben (Modell C, siehe internal/tenant) — Superadmin-Konten leben
|
||||
// dagegen mandantenuebergreifend in der Registry und sind ueber
|
||||
// SuperadminStore als eigener, First-Class-Typ modelliert, nicht als
|
||||
// tenant_id-NULL-Sonderfall in User.
|
||||
package user
|
||||
|
||||
import (
|
||||
"errors"
|
||||
"regexp"
|
||||
"time"
|
||||
)
|
||||
|
||||
type Status string
|
||||
|
||||
const (
|
||||
StatusActive Status = "active"
|
||||
StatusInactive Status = "inactive"
|
||||
)
|
||||
|
||||
type User struct {
|
||||
ID string
|
||||
Email string
|
||||
Name string
|
||||
Status Status
|
||||
CreatedAt time.Time
|
||||
UpdatedAt time.Time
|
||||
}
|
||||
|
||||
var emailPattern = regexp.MustCompile(`^[^\s@]+@[^\s@]+\.[^\s@]+$`)
|
||||
|
||||
var ErrInvalidEmail = errors.New("user: ungueltige E-Mail-Adresse")
|
||||
var ErrEmailTaken = errors.New("user: E-Mail-Adresse bereits vergeben")
|
||||
var ErrNotFound = errors.New("user: nicht gefunden")
|
||||
|
||||
func ValidateEmail(email string) error {
|
||||
if !emailPattern.MatchString(email) {
|
||||
return ErrInvalidEmail
|
||||
}
|
||||
return nil
|
||||
}
|
||||
@@ -0,0 +1,24 @@
|
||||
package user
|
||||
|
||||
import "testing"
|
||||
|
||||
func TestValidateEmail(t *testing.T) {
|
||||
cases := []struct {
|
||||
email string
|
||||
wantErr bool
|
||||
}{
|
||||
{"a@b.de", false},
|
||||
{"a.b+c@sub.example.com", false},
|
||||
{"", true},
|
||||
{"keine-email", true},
|
||||
{"a@b", true},
|
||||
{"@b.de", true},
|
||||
}
|
||||
|
||||
for _, c := range cases {
|
||||
err := ValidateEmail(c.email)
|
||||
if (err != nil) != c.wantErr {
|
||||
t.Errorf("ValidateEmail(%q) error = %v, wantErr %v", c.email, err, c.wantErr)
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,177 @@
|
||||
package webhook
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"context"
|
||||
"crypto/subtle"
|
||||
"encoding/hex"
|
||||
"fmt"
|
||||
"net/http"
|
||||
"time"
|
||||
|
||||
"github.com/jackc/pgx/v5"
|
||||
)
|
||||
|
||||
// DefaultMaxAttempts ist die konfigurierbare Obergrenze, ab der eine
|
||||
// Zustellung endgueltig als fehlgeschlagen gilt (Akzeptanzkriterium 2).
|
||||
const DefaultMaxAttempts = 5
|
||||
|
||||
// DefaultBaseBackoff ist die Basisdauer fuer exponentielles Backoff:
|
||||
// naechster Versuch nach BaseBackoff * 2^attempt (Akzeptanzkriterium 2).
|
||||
const DefaultBaseBackoff = 2 * time.Second
|
||||
|
||||
// Dispatcher liefert faellige Zustellungen aus. Konfigurierbar in Tests
|
||||
// (kleine BaseBackoff, kleine MaxAttempts), damit Retry/Backoff/Obergrenze
|
||||
// ohne minutenlange Wartezeit real durchlaufen werden koennen.
|
||||
type Dispatcher struct {
|
||||
pool pgxIface
|
||||
client *http.Client
|
||||
MaxAttempts int
|
||||
BaseBackoff time.Duration
|
||||
}
|
||||
|
||||
// pgxIface ist die schmale Teilmenge von *pgxpool.Pool, die der Dispatcher
|
||||
// braucht — als Interface, damit Tests keine echte Verbindung fuer reine
|
||||
// Signatur-/Backoff-Logik brauchen (wird hier aber durchgehend mit echten
|
||||
// Integrationstests gegen Postgres verwendet, siehe dispatcher_test.go).
|
||||
type pgxIface interface {
|
||||
Begin(ctx context.Context) (pgx.Tx, error)
|
||||
}
|
||||
|
||||
func NewDispatcher(pool pgxIface, client *http.Client) *Dispatcher {
|
||||
if client == nil {
|
||||
client = &http.Client{Timeout: 5 * time.Second}
|
||||
}
|
||||
return &Dispatcher{pool: pool, client: client, MaxAttempts: DefaultMaxAttempts, BaseBackoff: DefaultBaseBackoff}
|
||||
}
|
||||
|
||||
// backoffFor berechnet die Wartezeit vor dem naechsten Versuch: exponentiell
|
||||
// wachsend mit der Anzahl bereits unternommener Versuche.
|
||||
func (d *Dispatcher) backoffFor(attempt int) time.Duration {
|
||||
return d.BaseBackoff * time.Duration(1<<uint(attempt))
|
||||
}
|
||||
|
||||
// ProcessDue liefert ALLE derzeit faelligen Zustellungen aus — dasselbe
|
||||
// SELECT ... FOR UPDATE SKIP LOCKED-Muster wie
|
||||
// internal/tenant.Lifecycle.ProcessDueDeletions, damit mehrere Dispatcher-
|
||||
// Instanzen dieselbe Zustellung nie doppelt bearbeiten.
|
||||
func (d *Dispatcher) ProcessDue(ctx context.Context) (int, error) {
|
||||
tx, err := d.pool.Begin(ctx)
|
||||
if err != nil {
|
||||
return 0, fmt.Errorf("transaktion starten: %w", err)
|
||||
}
|
||||
defer func() { _ = tx.Rollback(ctx) }()
|
||||
|
||||
rows, err := tx.Query(ctx, `
|
||||
SELECT wd.id, wd.payload, wd.attempt, ws.target_url, ws.secret
|
||||
FROM webhook_deliveries wd
|
||||
JOIN webhook_subscriptions ws ON ws.id = wd.subscription_id
|
||||
WHERE wd.status = $1 AND wd.next_attempt_at <= now()
|
||||
FOR UPDATE OF wd SKIP LOCKED
|
||||
`, StatusPending)
|
||||
if err != nil {
|
||||
return 0, fmt.Errorf("faellige zustellungen abfragen: %w", err)
|
||||
}
|
||||
|
||||
var deliveries []delivery
|
||||
for rows.Next() {
|
||||
var del delivery
|
||||
if err := rows.Scan(&del.ID, &del.Payload, &del.Attempt, &del.TargetURL, &del.Secret); err != nil {
|
||||
rows.Close()
|
||||
return 0, fmt.Errorf("zustellung lesen: %w", err)
|
||||
}
|
||||
deliveries = append(deliveries, del)
|
||||
}
|
||||
rows.Close()
|
||||
if err := rows.Err(); err != nil {
|
||||
return 0, err
|
||||
}
|
||||
|
||||
for _, del := range deliveries {
|
||||
d.attemptOne(ctx, tx, del)
|
||||
}
|
||||
|
||||
if err := tx.Commit(ctx); err != nil {
|
||||
return 0, fmt.Errorf("transaktion committen: %w", err)
|
||||
}
|
||||
return len(deliveries), nil
|
||||
}
|
||||
|
||||
// attemptOne fuehrt GENAU EINEN Zustellversuch aus und aktualisiert den
|
||||
// Zustellungsdatensatz entsprechend — Erfolg (Akzeptanzkriterium 1/Pruefung
|
||||
// 1), erneuter Fehlversuch mit Backoff, oder endgueltiges Scheitern nach
|
||||
// DefaultMaxAttempts (Akzeptanzkriterium 2/Pruefung 2). Ein Fehler bei
|
||||
// GENAU EINER Zustellung darf die anderen in diesem Batch nicht verhindern.
|
||||
func (d *Dispatcher) attemptOne(ctx context.Context, tx pgx.Tx, del delivery) {
|
||||
signature := Sign(del.Secret, del.Payload)
|
||||
|
||||
req, err := http.NewRequestWithContext(ctx, http.MethodPost, del.TargetURL, bytes.NewReader(del.Payload))
|
||||
deliveryErr := err
|
||||
var statusCode int
|
||||
if err == nil {
|
||||
req.Header.Set("Content-Type", "application/json")
|
||||
req.Header.Set(SignatureHeader, signature)
|
||||
resp, err := d.client.Do(req)
|
||||
if err != nil {
|
||||
deliveryErr = err
|
||||
} else {
|
||||
statusCode = resp.StatusCode
|
||||
resp.Body.Close()
|
||||
if statusCode < 200 || statusCode >= 300 {
|
||||
deliveryErr = fmt.Errorf("unerwarteter statuscode %d", statusCode)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if deliveryErr == nil {
|
||||
_, _ = tx.Exec(ctx, `
|
||||
UPDATE webhook_deliveries SET status = $2, delivered_at = now(), attempt = attempt + 1
|
||||
WHERE id = $1
|
||||
`, del.ID, StatusDelivered)
|
||||
return
|
||||
}
|
||||
|
||||
nextAttempt := del.Attempt + 1
|
||||
if nextAttempt >= d.MaxAttempts {
|
||||
_, _ = tx.Exec(ctx, `
|
||||
UPDATE webhook_deliveries SET status = $2, attempt = $3, last_error = $4
|
||||
WHERE id = $1
|
||||
`, del.ID, StatusFailed, nextAttempt, deliveryErr.Error())
|
||||
return
|
||||
}
|
||||
|
||||
nextAttemptAt := time.Now().Add(d.backoffFor(nextAttempt))
|
||||
_, _ = tx.Exec(ctx, `
|
||||
UPDATE webhook_deliveries SET attempt = $2, next_attempt_at = $3, last_error = $4
|
||||
WHERE id = $1
|
||||
`, del.ID, nextAttempt, nextAttemptAt, deliveryErr.Error())
|
||||
}
|
||||
|
||||
// Run ruft ProcessDue in festen Abstaenden auf, bis ctx beendet wird —
|
||||
// dieselbe Konvention wie internal/tenant.Lifecycle.RunSweeper.
|
||||
func (d *Dispatcher) Run(ctx context.Context, interval time.Duration) {
|
||||
ticker := time.NewTicker(interval)
|
||||
defer ticker.Stop()
|
||||
for {
|
||||
select {
|
||||
case <-ctx.Done():
|
||||
return
|
||||
case <-ticker.C:
|
||||
_, _ = d.ProcessDue(ctx)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// VerifySignature prueft empfaengerseitig, ob signature zu payload und
|
||||
// secret passt — timing-safe (dasselbe Muster wie internal/audit.timingsafe),
|
||||
// damit ein Empfaenger die Authentizitaet einer Zustellung pruefen kann
|
||||
// (Akzeptanzkriterium 3).
|
||||
func VerifySignature(secret string, payload []byte, signature string) bool {
|
||||
expected := Sign(secret, payload)
|
||||
expectedBytes, err1 := hex.DecodeString(expected)
|
||||
gotBytes, err2 := hex.DecodeString(signature)
|
||||
if err1 != nil || err2 != nil {
|
||||
return false
|
||||
}
|
||||
return subtle.ConstantTimeCompare(expectedBytes, gotBytes) == 1
|
||||
}
|
||||
@@ -0,0 +1,131 @@
|
||||
// Package webhook implementiert Core API-07: eine zentrale Webhook-Registry
|
||||
// und Zustellungs-Engine fuer alle Fachmodule (DMS/Mail/Archive/Workflow/AI).
|
||||
// Module reichen Ereignisse EINMAL zur Zustellung ein und implementieren
|
||||
// selbst KEINE eigene Retry-/Signatur-Logik — das ist der zentrale Zweck
|
||||
// dieser Kachel ("Bewusst vermeiden: jedes Modul baut seine eigene
|
||||
// Webhook-Zustellungs-Engine"). Zustellung laeuft ueber dieselbe
|
||||
// Postgres-Jobqueue-Konvention (SELECT ... FOR UPDATE SKIP LOCKED) wie
|
||||
// internal/tenant.Lifecycle.ProcessDueDeletions (TEN-04) und
|
||||
// internal/notify.Dispatcher (CFG-02) — kein Redis/AMQP.
|
||||
package webhook
|
||||
|
||||
import (
|
||||
"context"
|
||||
"crypto/hmac"
|
||||
"crypto/sha256"
|
||||
"encoding/hex"
|
||||
"encoding/json"
|
||||
"fmt"
|
||||
"time"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
const (
|
||||
StatusPending = "pending"
|
||||
StatusDelivered = "delivered"
|
||||
StatusFailed = "failed"
|
||||
)
|
||||
|
||||
// Subscription ist EIN externer Abonnent fuer einen Ereignistyp
|
||||
// (Akzeptanzkriterium 1: Module registrieren Ereignistypen, externe
|
||||
// Abonnenten registrieren Ziel-URLs — dieses Paket modelliert die
|
||||
// Abonnenten-Seite; welche Ereignistypen ein Modul anbietet, ist bewusst
|
||||
// NICHT Teil dieser Kachel).
|
||||
type Subscription struct {
|
||||
ID string
|
||||
EventType string
|
||||
TargetURL string
|
||||
Secret string
|
||||
CreatedAt time.Time
|
||||
}
|
||||
|
||||
// Store persistiert Abonnements und Zustellversuche.
|
||||
type Store struct {
|
||||
pool *pgxpool.Pool
|
||||
}
|
||||
|
||||
func NewStore(pool *pgxpool.Pool) *Store {
|
||||
return &Store{pool: pool}
|
||||
}
|
||||
|
||||
// Subscribe registriert einen Abonnenten fuer einen Ereignistyp. secret wird
|
||||
// spaeter zur HMAC-Signierung jeder Zustellung an diesen Abonnenten
|
||||
// verwendet (Akzeptanzkriterium 3).
|
||||
func (s *Store) Subscribe(ctx context.Context, eventType, targetURL, secret string) (Subscription, error) {
|
||||
var sub Subscription
|
||||
sub.EventType, sub.TargetURL, sub.Secret = eventType, targetURL, secret
|
||||
row := s.pool.QueryRow(ctx, `
|
||||
INSERT INTO webhook_subscriptions (event_type, target_url, secret)
|
||||
VALUES ($1, $2, $3)
|
||||
RETURNING id, created_at
|
||||
`, eventType, targetURL, secret)
|
||||
if err := row.Scan(&sub.ID, &sub.CreatedAt); err != nil {
|
||||
return Subscription{}, fmt.Errorf("abonnement anlegen: %w", err)
|
||||
}
|
||||
return sub, nil
|
||||
}
|
||||
|
||||
// Enqueue reicht EIN Ereignis zur Zustellung an ALLE Abonnenten des
|
||||
// angegebenen Ereignistyps ein — dies ist die EINZIGE Schnittstelle, die
|
||||
// ein Fachmodul braucht (Akzeptanzkriterium 1). Jeder Abonnent erhaelt
|
||||
// einen eigenen, unabhaengigen Zustellversuch-Datensatz.
|
||||
func (s *Store) Enqueue(ctx context.Context, eventType string, payload any) (int, error) {
|
||||
payloadJSON, err := json.Marshal(payload)
|
||||
if err != nil {
|
||||
return 0, fmt.Errorf("payload serialisieren: %w", err)
|
||||
}
|
||||
|
||||
rows, err := s.pool.Query(ctx, `
|
||||
SELECT id FROM webhook_subscriptions WHERE event_type = $1
|
||||
`, eventType)
|
||||
if err != nil {
|
||||
return 0, fmt.Errorf("abonnenten ermitteln: %w", err)
|
||||
}
|
||||
var subscriptionIDs []string
|
||||
for rows.Next() {
|
||||
var id string
|
||||
if err := rows.Scan(&id); err != nil {
|
||||
rows.Close()
|
||||
return 0, fmt.Errorf("abonnent lesen: %w", err)
|
||||
}
|
||||
subscriptionIDs = append(subscriptionIDs, id)
|
||||
}
|
||||
rows.Close()
|
||||
if err := rows.Err(); err != nil {
|
||||
return 0, err
|
||||
}
|
||||
|
||||
for _, subID := range subscriptionIDs {
|
||||
if _, err := s.pool.Exec(ctx, `
|
||||
INSERT INTO webhook_deliveries (subscription_id, event_type, payload, status, next_attempt_at)
|
||||
VALUES ($1, $2, $3, $4, now())
|
||||
`, subID, eventType, payloadJSON, StatusPending); err != nil {
|
||||
return 0, fmt.Errorf("zustellung einreihen: %w", err)
|
||||
}
|
||||
}
|
||||
return len(subscriptionIDs), nil
|
||||
}
|
||||
|
||||
// delivery ist ein interner Datensatz fuer EINEN Zustellversuch, inklusive
|
||||
// der zugehoerigen Abonnentendaten (per JOIN geladen).
|
||||
type delivery struct {
|
||||
ID string
|
||||
TargetURL string
|
||||
Secret string
|
||||
Payload []byte
|
||||
Attempt int
|
||||
}
|
||||
|
||||
// Sign berechnet die HMAC-SHA256-Signatur des Payloads (Akzeptanzkriterium
|
||||
// 3) — hex-kodiert, damit sie problemlos als HTTP-Header uebertragen werden
|
||||
// kann.
|
||||
func Sign(secret string, payload []byte) string {
|
||||
mac := hmac.New(sha256.New, []byte(secret))
|
||||
mac.Write(payload)
|
||||
return hex.EncodeToString(mac.Sum(nil))
|
||||
}
|
||||
|
||||
// SignatureHeader ist der HTTP-Header, unter dem die Signatur uebertragen
|
||||
// wird — dokumentierter Vertrag fuer Empfaenger (Akzeptanzkriterium 3).
|
||||
const SignatureHeader = "X-Nexarch-Signature-256"
|
||||
@@ -0,0 +1,205 @@
|
||||
package webhook
|
||||
|
||||
import (
|
||||
"context"
|
||||
"encoding/json"
|
||||
"io"
|
||||
"net/http"
|
||||
"net/http/httptest"
|
||||
"os"
|
||||
"sync/atomic"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/jackc/pgx/v5/pgxpool"
|
||||
)
|
||||
|
||||
func setupTest(t *testing.T) (*Store, *pgxpool.Pool, func()) {
|
||||
t.Helper()
|
||||
adminDSN := os.Getenv("TEST_ADMIN_DSN")
|
||||
if adminDSN == "" {
|
||||
t.Skip("TEST_ADMIN_DSN nicht gesetzt, Integrationstest uebersprungen")
|
||||
}
|
||||
ctx := context.Background()
|
||||
|
||||
pool, err := pgxpool.New(ctx, adminDSN)
|
||||
if err != nil {
|
||||
t.Fatalf("pool: %v", err)
|
||||
}
|
||||
if _, err := pool.Exec(ctx, `
|
||||
CREATE TABLE IF NOT EXISTS webhook_subscriptions (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(), event_type TEXT NOT NULL,
|
||||
target_url TEXT NOT NULL, secret TEXT NOT NULL, created_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
CREATE TABLE IF NOT EXISTS webhook_deliveries (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(), subscription_id UUID NOT NULL REFERENCES webhook_subscriptions(id),
|
||||
event_type TEXT NOT NULL, payload JSONB NOT NULL, status TEXT NOT NULL DEFAULT 'pending',
|
||||
attempt INT NOT NULL DEFAULT 0, next_attempt_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
last_error TEXT, created_at TIMESTAMPTZ NOT NULL DEFAULT now(), delivered_at TIMESTAMPTZ
|
||||
);
|
||||
`); err != nil {
|
||||
t.Fatalf("schema: %v", err)
|
||||
}
|
||||
|
||||
cleanup := func() { pool.Close() }
|
||||
return NewStore(pool), pool, cleanup
|
||||
}
|
||||
|
||||
func uniqueEventType(prefix string) string {
|
||||
return prefix + "-" + time.Now().Format("150405.000000000")
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 1 + Pruefung 1: ein Modul reicht ein Ereignis ein
|
||||
// (Enqueue) ohne eigene Zustellungslogik, der zentrale Dispatcher liefert
|
||||
// zuverlaessig aus.
|
||||
func TestEnqueueAndProcessDue_DeliversSuccessfully(t *testing.T) {
|
||||
store, pool, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
var receivedBody []byte
|
||||
var receivedSignature string
|
||||
target := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
|
||||
body, _ := io.ReadAll(r.Body)
|
||||
receivedBody = body
|
||||
receivedSignature = r.Header.Get(SignatureHeader)
|
||||
w.WriteHeader(http.StatusOK)
|
||||
}))
|
||||
defer target.Close()
|
||||
|
||||
eventType := uniqueEventType("dms.file.created")
|
||||
sub, err := store.Subscribe(ctx, eventType, target.URL, "geheimes-secret")
|
||||
if err != nil {
|
||||
t.Fatalf("subscribe: %v", err)
|
||||
}
|
||||
|
||||
n, err := store.Enqueue(ctx, eventType, map[string]string{"file_id": "42"})
|
||||
if err != nil {
|
||||
t.Fatalf("enqueue: %v", err)
|
||||
}
|
||||
if n != 1 {
|
||||
t.Fatalf("erwartet 1 eingereihte zustellung, habe %d", n)
|
||||
}
|
||||
|
||||
dispatcher := NewDispatcher(pool, target.Client())
|
||||
processed, err := dispatcher.ProcessDue(ctx)
|
||||
if err != nil {
|
||||
t.Fatalf("processdue: %v", err)
|
||||
}
|
||||
if processed != 1 {
|
||||
t.Fatalf("erwartet 1 verarbeitete zustellung, habe %d", processed)
|
||||
}
|
||||
|
||||
var status string
|
||||
if err := pool.QueryRow(ctx, `SELECT status FROM webhook_deliveries WHERE subscription_id = $1`, sub.ID).Scan(&status); err != nil {
|
||||
t.Fatalf("status lesen: %v", err)
|
||||
}
|
||||
if status != StatusDelivered {
|
||||
t.Fatalf("status = %q, want %q", status, StatusDelivered)
|
||||
}
|
||||
|
||||
// Postgres' JSONB-Spalte kann die Byte-Repraesentation des Payloads
|
||||
// gegenueber dem urspruenglichen json.Marshal kanonisieren (z.B.
|
||||
// Leerzeichen) — das ist unschaedlich, denn der Dispatcher signiert
|
||||
// IMMER exakt die Bytes, die er auch sendet. Die Pruefung vergleicht
|
||||
// deshalb Signatur gegen tatsaechlich empfangene Bytes (Selbstkonsistenz),
|
||||
// nicht gegen eine unabhaengig neu marshalte Referenz.
|
||||
if !VerifySignature("geheimes-secret", receivedBody, receivedSignature) {
|
||||
t.Fatalf("empfangene signatur %q passt nicht zum empfangenen payload %q", receivedSignature, receivedBody)
|
||||
}
|
||||
var decoded map[string]string
|
||||
if err := json.Unmarshal(receivedBody, &decoded); err != nil {
|
||||
t.Fatalf("empfangener payload nicht als json lesbar: %v", err)
|
||||
}
|
||||
if decoded["file_id"] != "42" {
|
||||
t.Fatalf("empfangener payload = %v, want file_id=42", decoded)
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 2 + Pruefung 2: fehlschlagendes Ziel loest Retry mit
|
||||
// wachsendem Backoff aus und endet nach der konfigurierten Obergrenze in
|
||||
// "failed".
|
||||
func TestProcessDue_RetriesWithBackoffThenMarksFailed(t *testing.T) {
|
||||
store, pool, cleanup := setupTest(t)
|
||||
defer cleanup()
|
||||
ctx := context.Background()
|
||||
|
||||
var callCount int32
|
||||
target := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
|
||||
atomic.AddInt32(&callCount, 1)
|
||||
w.WriteHeader(http.StatusInternalServerError)
|
||||
}))
|
||||
defer target.Close()
|
||||
|
||||
eventType := uniqueEventType("mail.send.failed")
|
||||
sub, err := store.Subscribe(ctx, eventType, target.URL, "secret")
|
||||
if err != nil {
|
||||
t.Fatalf("subscribe: %v", err)
|
||||
}
|
||||
if _, err := store.Enqueue(ctx, eventType, map[string]string{"x": "y"}); err != nil {
|
||||
t.Fatalf("enqueue: %v", err)
|
||||
}
|
||||
|
||||
dispatcher := NewDispatcher(pool, target.Client())
|
||||
dispatcher.MaxAttempts = 2
|
||||
dispatcher.BaseBackoff = 30 * time.Millisecond
|
||||
|
||||
// 1. Versuch: schlaegt fehl, ist aber noch nicht die Obergrenze.
|
||||
if _, err := dispatcher.ProcessDue(ctx); err != nil {
|
||||
t.Fatalf("processdue 1: %v", err)
|
||||
}
|
||||
var status string
|
||||
var attempt int
|
||||
if err := pool.QueryRow(ctx, `SELECT status, attempt FROM webhook_deliveries WHERE subscription_id = $1`, sub.ID).Scan(&status, &attempt); err != nil {
|
||||
t.Fatalf("status lesen 1: %v", err)
|
||||
}
|
||||
if status != StatusPending || attempt != 1 {
|
||||
t.Fatalf("nach 1. fehlschlag: status=%q attempt=%d, want pending/1", status, attempt)
|
||||
}
|
||||
|
||||
// Sofort erneut verarbeiten: Backoff ist noch nicht abgelaufen -> nichts faellig.
|
||||
processedTooEarly, err := dispatcher.ProcessDue(ctx)
|
||||
if err != nil {
|
||||
t.Fatalf("processdue (zu frueh): %v", err)
|
||||
}
|
||||
if processedTooEarly != 0 {
|
||||
t.Fatal("erwartet 0 verarbeitete zustellungen, solange backoff nicht abgelaufen ist")
|
||||
}
|
||||
|
||||
time.Sleep(dispatcher.backoffFor(1) + 20*time.Millisecond)
|
||||
|
||||
// 2. Versuch: erreicht MaxAttempts=2 -> endgueltig fehlgeschlagen.
|
||||
if _, err := dispatcher.ProcessDue(ctx); err != nil {
|
||||
t.Fatalf("processdue 2: %v", err)
|
||||
}
|
||||
if err := pool.QueryRow(ctx, `SELECT status, attempt FROM webhook_deliveries WHERE subscription_id = $1`, sub.ID).Scan(&status, &attempt); err != nil {
|
||||
t.Fatalf("status lesen 2: %v", err)
|
||||
}
|
||||
if status != StatusFailed || attempt != 2 {
|
||||
t.Fatalf("nach 2. fehlschlag: status=%q attempt=%d, want failed/2", status, attempt)
|
||||
}
|
||||
if atomic.LoadInt32(&callCount) != 2 {
|
||||
t.Fatalf("erwartet genau 2 tatsaechliche zustellversuche, habe %d", callCount)
|
||||
}
|
||||
}
|
||||
|
||||
// Akzeptanzkriterium 3 + Pruefung 3: Signaturpruefung erkennt eine
|
||||
// manipulierte Payload zuverlaessig.
|
||||
func TestVerifySignature_DetectsTamperedPayload(t *testing.T) {
|
||||
secret := "geteiltes-geheimnis"
|
||||
payload := []byte(`{"file_id":"42"}`)
|
||||
signature := Sign(secret, payload)
|
||||
|
||||
if !VerifySignature(secret, payload, signature) {
|
||||
t.Fatal("erwartet gueltige signatur fuer unveraenderte payload")
|
||||
}
|
||||
|
||||
tampered := []byte(`{"file_id":"99"}`)
|
||||
if VerifySignature(secret, tampered, signature) {
|
||||
t.Fatal("erwartet ungueltige signatur fuer manipulierte payload")
|
||||
}
|
||||
|
||||
if VerifySignature("falsches-secret", payload, signature) {
|
||||
t.Fatal("erwartet ungueltige signatur bei falschem secret")
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1 @@
|
||||
DROP TABLE IF EXISTS superadmins;
|
||||
@@ -0,0 +1,14 @@
|
||||
-- Superadmin-Konten arbeiten mandantenuebergreifend und leben deshalb in der
|
||||
-- Control-Plane-Registry (siehe TEN-01), nicht in einer Tenant-Datenbank.
|
||||
-- Das bildet "Superadmin ohne Tenant" strukturell als First-Class-Zustand ab,
|
||||
-- statt ihn als Sonderfall in der Tenant-users-Tabelle zu behandeln
|
||||
-- (IAM-01, siehe core-kanban/tickets/IAM-01.md — bekannte Fehler vermeiden).
|
||||
-- E-Mail-Eindeutigkeit ist hier global, da die Registry-DB einmalig existiert.
|
||||
CREATE TABLE superadmins (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
email TEXT NOT NULL UNIQUE,
|
||||
name TEXT NOT NULL,
|
||||
status TEXT NOT NULL DEFAULT 'active',
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
@@ -0,0 +1,2 @@
|
||||
DROP TABLE webhook_deliveries;
|
||||
DROP TABLE webhook_subscriptions;
|
||||
@@ -0,0 +1,27 @@
|
||||
-- Zentrale Webhook-Registry & Zustellung (API-07, siehe
|
||||
-- core-kanban/tickets/API-07.md) — Postgres-basierte Jobqueue, kein
|
||||
-- Redis/AMQP (Ticket-Vorgabe).
|
||||
CREATE TABLE webhook_subscriptions (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
event_type TEXT NOT NULL,
|
||||
target_url TEXT NOT NULL,
|
||||
secret TEXT NOT NULL,
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
|
||||
CREATE INDEX webhook_subscriptions_event_type_idx ON webhook_subscriptions (event_type);
|
||||
|
||||
CREATE TABLE webhook_deliveries (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
subscription_id UUID NOT NULL REFERENCES webhook_subscriptions(id),
|
||||
event_type TEXT NOT NULL,
|
||||
payload JSONB NOT NULL,
|
||||
status TEXT NOT NULL DEFAULT 'pending', -- pending|delivered|failed
|
||||
attempt INT NOT NULL DEFAULT 0,
|
||||
next_attempt_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
last_error TEXT,
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
delivered_at TIMESTAMPTZ
|
||||
);
|
||||
|
||||
CREATE INDEX webhook_deliveries_due_idx ON webhook_deliveries (status, next_attempt_at);
|
||||
@@ -1,2 +0,0 @@
|
||||
DROP TABLE IF EXISTS config_value_history;
|
||||
DROP TABLE IF EXISTS config_values;
|
||||
@@ -1,23 +0,0 @@
|
||||
-- Zentraler Konfigurationsdienst (CFG-01, siehe core-kanban/tickets/CFG-01.md).
|
||||
-- scope = 'global' fuer globale Defaults, sonst der Tenant-Slug. config_values
|
||||
-- haelt den AKTUELLEN Stand je (key, scope); config_value_history haelt JEDE
|
||||
-- Aenderung fest (Akzeptanzkriterium 2: versioniert nachvollziehbar).
|
||||
CREATE TABLE config_values (
|
||||
key TEXT NOT NULL,
|
||||
scope TEXT NOT NULL CHECK (scope <> ''),
|
||||
value TEXT NOT NULL,
|
||||
version INT NOT NULL,
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
PRIMARY KEY (key, scope)
|
||||
);
|
||||
|
||||
CREATE TABLE config_value_history (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
key TEXT NOT NULL,
|
||||
scope TEXT NOT NULL,
|
||||
value TEXT NOT NULL,
|
||||
version INT NOT NULL,
|
||||
changed_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
|
||||
CREATE INDEX config_value_history_key_scope_idx ON config_value_history (key, scope, version);
|
||||
@@ -1 +0,0 @@
|
||||
DROP TABLE IF EXISTS feature_flags;
|
||||
@@ -1,10 +0,0 @@
|
||||
-- Feature-Flags zentral je Mandant/Zielgruppe (LIC-02, siehe core-kanban/tickets/LIC-02.md).
|
||||
-- Lebt in der Registry-DB, nicht pro Tenant-Datenbank — Flags sind eine
|
||||
-- Core-weite Konfiguration, keine Mandanten-Geschaeftsdaten.
|
||||
CREATE TABLE feature_flags (
|
||||
key TEXT PRIMARY KEY,
|
||||
enabled BOOLEAN NOT NULL DEFAULT false,
|
||||
rollout_percentage INT NOT NULL DEFAULT 0 CHECK (rollout_percentage BETWEEN 0 AND 100),
|
||||
target_tenant_slugs TEXT[] NOT NULL DEFAULT '{}',
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
@@ -1,2 +0,0 @@
|
||||
DROP TABLE IF EXISTS module_credentials;
|
||||
DROP TABLE IF EXISTS modules;
|
||||
@@ -1,18 +0,0 @@
|
||||
-- Modul-Registry & Aktivierungspruefung (API-02, siehe core-kanban/tickets/API-02.md).
|
||||
CREATE TABLE modules (
|
||||
name TEXT PRIMARY KEY,
|
||||
version TEXT NOT NULL CHECK (version <> ''),
|
||||
required_flags TEXT[] NOT NULL DEFAULT '{}',
|
||||
registered_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
|
||||
-- Service-Credential je Modul-Instanz, bei Provisionierung ausgestellt
|
||||
-- (Akzeptanzkriterium 4). secret_hash enthaelt NIEMALS das Secret im
|
||||
-- Klartext, nur dessen SHA-256-Hash (Timing-safe-Vergleich beim Login,
|
||||
-- Referenzmuster siehe AUD-02).
|
||||
CREATE TABLE module_credentials (
|
||||
module_name TEXT PRIMARY KEY REFERENCES modules(name),
|
||||
client_id TEXT NOT NULL UNIQUE,
|
||||
secret_hash BYTEA NOT NULL,
|
||||
issued_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
@@ -1 +0,0 @@
|
||||
DROP TABLE IF EXISTS notification_jobs;
|
||||
@@ -1,20 +0,0 @@
|
||||
-- Benachrichtigungs-Dispatcher-Warteschlange (CFG-02, siehe
|
||||
-- core-kanban/tickets/CFG-02.md). Postgres-basiert statt Redis/AMQP
|
||||
-- (Projekt-Konvention, siehe nexarch-state.json techstack.job_queue) —
|
||||
-- Zeilen ueberleben einen Neustart des Dispatcher-Prozesses unveraendert
|
||||
-- (Akzeptanzkriterium 3).
|
||||
CREATE TABLE notification_jobs (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
channel TEXT NOT NULL,
|
||||
recipient TEXT NOT NULL,
|
||||
payload JSONB NOT NULL DEFAULT '{}'::jsonb,
|
||||
status TEXT NOT NULL DEFAULT 'pending' CHECK (status IN ('pending', 'sent', 'failed')),
|
||||
attempts INT NOT NULL DEFAULT 0,
|
||||
max_attempts INT NOT NULL DEFAULT 5,
|
||||
next_attempt_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
last_error TEXT,
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
|
||||
CREATE INDEX notification_jobs_due_idx ON notification_jobs (status, next_attempt_at);
|
||||
@@ -1,2 +0,0 @@
|
||||
DROP TABLE alert_debounce_state;
|
||||
DROP TABLE alert_rules;
|
||||
@@ -1,24 +0,0 @@
|
||||
-- OPS-05: Schwellwert-Regeln fuer Alerting auf den aus OPS-03 aggregierten
|
||||
-- Metriken. Lebt wie config_values/notification_jobs (CFG-01/02) in der
|
||||
-- zentralen Registry-DB — modulübergreifende Betriebskonfiguration, keine
|
||||
-- Mandanten-Geschaeftsdaten.
|
||||
CREATE TABLE alert_rules (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
metric_name TEXT NOT NULL,
|
||||
comparison TEXT NOT NULL CHECK (comparison IN ('gt', 'lt')),
|
||||
threshold DOUBLE PRECISION NOT NULL,
|
||||
label_filters JSONB NOT NULL DEFAULT '{}'::jsonb,
|
||||
recipient TEXT NOT NULL,
|
||||
description TEXT NOT NULL DEFAULT '',
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
|
||||
-- Haelt fest, wann eine Regel zuletzt tatsaechlich einen Alarm ausgeloest
|
||||
-- hat (Akzeptanzkriterium 3: Drosselung wiederholter Alarmierung fuer
|
||||
-- denselben anhaltenden Zustand). rule_key kombiniert Regel-ID mit den
|
||||
-- tatsaechlichen Label-Werten der ausloesenden Zeitreihe, damit dieselbe
|
||||
-- Regel fuer unterschiedliche Tenants/Module unabhaengig gedrosselt wird.
|
||||
CREATE TABLE alert_debounce_state (
|
||||
rule_key TEXT PRIMARY KEY,
|
||||
last_fired_at TIMESTAMPTZ NOT NULL
|
||||
);
|
||||
@@ -1 +0,0 @@
|
||||
DROP TABLE metrics_sources;
|
||||
@@ -1,7 +0,0 @@
|
||||
-- Metrics-Aggregation ueber Module hinweg (OPS-03, siehe
|
||||
-- core-kanban/tickets/OPS-03.md) — welches Modul liefert seine Kennzahlen
|
||||
-- unter welcher /metrics-URL.
|
||||
CREATE TABLE metrics_sources (
|
||||
module_name TEXT PRIMARY KEY,
|
||||
metrics_url TEXT NOT NULL
|
||||
);
|
||||
@@ -0,0 +1 @@
|
||||
DROP TABLE IF EXISTS users;
|
||||
@@ -0,0 +1,16 @@
|
||||
-- Benutzer-Datenmodell (IAM-01, siehe core-kanban/tickets/IAM-01.md).
|
||||
-- Diese Migration laeuft in der DB EINES Mandanten (Modell C, siehe TEN-01) —
|
||||
-- die Tenant-Zugehoerigkeit ist implizit durch die Datenbankverbindung
|
||||
-- gegeben, es gibt daher bewusst KEINE tenant_id-Spalte.
|
||||
-- E-Mail-Eindeutigkeit ist hier tenant-scoped: der UNIQUE-Constraint gilt
|
||||
-- nur innerhalb dieser einen Tenant-Datenbank.
|
||||
CREATE EXTENSION IF NOT EXISTS pgcrypto;
|
||||
|
||||
CREATE TABLE users (
|
||||
id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
|
||||
email TEXT NOT NULL UNIQUE,
|
||||
name TEXT NOT NULL,
|
||||
status TEXT NOT NULL DEFAULT 'active',
|
||||
created_at TIMESTAMPTZ NOT NULL DEFAULT now(),
|
||||
updated_at TIMESTAMPTZ NOT NULL DEFAULT now()
|
||||
);
|
||||
@@ -0,0 +1 @@
|
||||
ALTER TABLE users DROP COLUMN password_hash;
|
||||
@@ -0,0 +1,3 @@
|
||||
-- Passwort-Hash-Spalte fuer Login (IAM-02, siehe core-kanban/tickets/IAM-02.md).
|
||||
-- Enthaelt AUSSCHLIESSLICH den bcrypt-Hash, niemals das Klartext-Passwort.
|
||||
ALTER TABLE users ADD COLUMN password_hash TEXT NOT NULL DEFAULT '';
|
||||
@@ -1,11 +1,4 @@
|
||||
#!/usr/bin/env bash
|
||||
# Setzt die nexarch-Testumgebung zurueck: loescht die geteilte
|
||||
# Registry-Tabelle "tenants" in der postgres-Wartungsdatenbank sowie alle
|
||||
# tenant_*-Datenbanken. Noetig, weil verschiedene Feature-Branches
|
||||
# unterschiedliche Registry-Schemata erwarten, aber dieselbe physische
|
||||
# Postgres-Instanz auf dem Testhost teilen (siehe [[project-nexarch-test-infra]]).
|
||||
#
|
||||
# Aufruf: NEXARCH_TEST_DB_PASSWORD=... ./scripts/reset-test-env.sh
|
||||
set -euo pipefail
|
||||
|
||||
PASS="${NEXARCH_TEST_DB_PASSWORD:?Setze NEXARCH_TEST_DB_PASSWORD vor dem Aufruf}"
|
||||
@@ -14,9 +7,6 @@ ROLE="nexarch_test"
|
||||
export PGPASSWORD="$PASS"
|
||||
|
||||
psql -h localhost -U "$ROLE" -d postgres -v ON_ERROR_STOP=1 -c "DROP TABLE IF EXISTS tenants CASCADE;"
|
||||
psql -h localhost -U "$ROLE" -d postgres -v ON_ERROR_STOP=1 -c "DROP TABLE IF EXISTS audit_events CASCADE;"
|
||||
psql -h localhost -U "$ROLE" -d postgres -v ON_ERROR_STOP=1 -c "DROP TABLE IF EXISTS config_value_history CASCADE;"
|
||||
psql -h localhost -U "$ROLE" -d postgres -v ON_ERROR_STOP=1 -c "DROP TABLE IF EXISTS config_values CASCADE;"
|
||||
|
||||
dbs=$(psql -h localhost -U "$ROLE" -d postgres -tAc "SELECT datname FROM pg_database WHERE datname LIKE 'tenant\_%' ESCAPE '\'")
|
||||
for db in $dbs; do
|
||||
|
||||
@@ -1,10 +1,4 @@
|
||||
#!/usr/bin/env bash
|
||||
# Ein-Kommando-Pruefung fuer den aktuellen Code-Stand auf dem Testhost:
|
||||
# Registry+Tenant-DBs zuruecksetzen, dann build/vet/test in einem Rutsch.
|
||||
# -p 1 ist Pflicht, da mehrere Pakete dieselbe physische Registry-Tabelle auf
|
||||
# dem Testhost teilen (siehe [[project-nexarch-test-infra]]).
|
||||
#
|
||||
# Aufruf: NEXARCH_TEST_DB_PASSWORD=... ./scripts/run-checks.sh
|
||||
set -euo pipefail
|
||||
|
||||
PASS="${NEXARCH_TEST_DB_PASSWORD:?Setze NEXARCH_TEST_DB_PASSWORD vor dem Aufruf}"
|
||||
|
||||
Reference in New Issue
Block a user