Publish ITWorx Pulse source
Public source validation / validate (push) Failing after 3m8s

This commit is contained in:
ITWorx Pulse release export
2026-09-03 02:09:19 +02:00
commit bd774932d5
614 changed files with 77116 additions and 0 deletions
+111
View File
@@ -0,0 +1,111 @@
package alertdefaults
import (
"bytes"
"context"
"embed"
"encoding/json"
"errors"
"fmt"
"github.com/itworx/pulse/internal/alert"
"github.com/itworx/pulse/internal/metriccatalog"
)
//go:embed seed.json
var seedFS embed.FS
type seedDocument struct {
SchemaVersion int `json:"schemaVersion"`
Rules []alert.Document `json:"rules"`
}
type RuleStore interface {
Create(context.Context, string, alert.Document, string) (alert.Rule, alert.Version, error)
Get(context.Context, string) (alert.Rule, error)
// Update is optional for the seed: stores that also implement it allow the
// seed to refresh an implementation-owned default that has never been edited
// (revision 1) when a newer seed corrects it. See Seed.
}
// RuleUpdater is implemented by stores that support versioned updates. The seed
// uses it only for rules that are still at revision 1 (created by the seed and
// never changed by an operator), so operator edits are never overwritten.
type RuleUpdater interface {
Update(ctx context.Context, id, actor string, expected int64, document alert.Document, changeSummary string) (alert.Rule, error)
}
type Report struct {
Added int
Existing int
Refreshed int
}
func Load(registry metriccatalog.Registry) ([]alert.Document, error) {
raw, err := seedFS.ReadFile("seed.json")
if err != nil {
return nil, fmt.Errorf("read embedded alert defaults: %w", err)
}
decoder := json.NewDecoder(bytes.NewReader(raw))
decoder.DisallowUnknownFields()
var bundle seedDocument
if err := decoder.Decode(&bundle); err != nil {
return nil, fmt.Errorf("decode embedded alert defaults: %w", err)
}
if bundle.SchemaVersion != 1 || len(bundle.Rules) == 0 {
return nil, errors.New("embedded alert defaults have an invalid bundle")
}
seen := make(map[string]struct{}, len(bundle.Rules))
for _, rule := range bundle.Rules {
if _, ok := seen[rule.ID]; ok {
return nil, fmt.Errorf("embedded alert defaults contain duplicate rule %q", rule.ID)
}
seen[rule.ID] = struct{}{}
if err := rule.Validate(registry); err != nil {
return nil, fmt.Errorf("validate embedded alert default %q: %w", rule.ID, err)
}
}
return bundle.Rules, nil
}
func Seed(ctx context.Context, store RuleStore, registry metriccatalog.Registry, actor string) (Report, error) {
if store == nil {
return Report{}, alert.ErrUnavailable
}
rules, err := Load(registry)
if err != nil {
return Report{}, err
}
report := Report{}
for _, document := range rules {
if _, _, err := store.Create(ctx, actor, document, "system default seed v1"); err == nil {
report.Added++
} else if errors.Is(err, alert.ErrConflict) {
existing, getErr := store.Get(ctx, document.ID)
if getErr != nil {
return Report{}, fmt.Errorf("verify existing alert default %q: %w", document.ID, getErr)
}
if updater, ok := store.(RuleUpdater); ok && existing.Revision == 1 && !sameDocument(existing.Document, document) {
// The stored default is untouched since the seed created it, but the
// seed itself changed (for example a corrected metric binding). Refresh
// it through the normal versioned update path so history is kept.
if _, updErr := updater.Update(ctx, document.ID, actor, existing.Revision, document, "system default seed v1 refresh"); updErr != nil {
return Report{}, fmt.Errorf("refresh alert default %q: %w", document.ID, updErr)
}
report.Refreshed++
continue
}
report.Existing++
} else {
return Report{}, fmt.Errorf("seed alert default %q: %w", document.ID, err)
}
}
return report, nil
}
// sameDocument compares two rule documents by their canonical JSON encoding.
func sameDocument(a, b alert.Document) bool {
left, errA := json.Marshal(a)
right, errB := json.Marshal(b)
return errA == nil && errB == nil && bytes.Equal(left, right)
}
+172
View File
@@ -0,0 +1,172 @@
{
"schemaVersion": 1,
"rules": [
{
"schemaVersion": 1,
"id": "71111111-1111-4111-8111-111111111111",
"name": "Monitoringbron levert geen recente gegevens",
"enabled": true,
"severity": "degraded",
"scope": {
"entityType": "data-source",
"required": true
},
"condition": {
"inputType": "datasource-health",
"operator": "==",
"threshold": "stale",
"windowSeconds": 120
},
"evaluationIntervalSeconds": 30,
"pendingSeconds": 120,
"resolveSeconds": 60,
"unknownBehavior": "become-unknown",
"groupBy": [
"source"
],
"suppressWhen": [],
"message": {
"titleKey": "alerts.datasourceStale.title",
"bodyKey": "alerts.datasourceStale.body"
}
},
{
"schemaVersion": 1,
"id": "81111111-1111-4111-8111-111111111111",
"name": "Container bevindt zich in een herstartlus",
"enabled": true,
"severity": "degraded",
"scope": {
"entityType": "container",
"excludeIntentionalStopped": true
},
"condition": {
"inputType": "event",
"operator": ">=",
"threshold": 3,
"aggregation": "count",
"windowSeconds": 900
},
"evaluationIntervalSeconds": 30,
"pendingSeconds": 0,
"resolveSeconds": 900,
"unknownBehavior": "become-unknown",
"groupBy": [
"container",
"application"
],
"suppressWhen": [
"host.unreachable"
],
"message": {
"titleKey": "alerts.containerRestartLoop.title",
"bodyKey": "alerts.containerRestartLoop.body"
}
},
{
"schemaVersion": 1,
"id": "91111111-1111-4111-8111-111111111111",
"name": "Disktemperatuur te hoog",
"enabled": true,
"severity": "degraded",
"scope": {
"entityType": "disk"
},
"condition": {
"inputType": "metric",
"metric": "storage.disk.temperature.maximum",
"operator": ">=",
"threshold": 50,
"recoveryThreshold": 46,
"aggregation": "max",
"windowSeconds": 300
},
"evaluationIntervalSeconds": 30,
"pendingSeconds": 300,
"resolveSeconds": 300,
"unknownBehavior": "retain-firing-as-unknown",
"groupBy": [
"disk",
"server"
],
"suppressWhen": [
"host.unreachable",
"storage.source.unavailable"
],
"message": {
"titleKey": "alerts.diskTemperature.title",
"bodyKey": "alerts.diskTemperature.body"
}
},
{
"schemaVersion": 1,
"id": "a1111111-1111-4111-8111-111111111111",
"name": "Service is niet bereikbaar",
"enabled": true,
"severity": "degraded",
"scope": {
"entityType": "service",
"critical": true
},
"condition": {
"inputType": "metric",
"metric": "service.availability.minimum",
"operator": "<",
"threshold": 1,
"aggregation": "min",
"windowSeconds": 90
},
"evaluationIntervalSeconds": 30,
"pendingSeconds": 90,
"resolveSeconds": 60,
"unknownBehavior": "become-unknown",
"groupBy": [
"service",
"application"
],
"suppressWhen": [
"host.unreachable",
"network.gateway.unreachable",
"dns.unavailable"
],
"message": {
"titleKey": "alerts.serviceUnavailable.title",
"bodyKey": "alerts.serviceUnavailable.body"
}
},
{
"schemaVersion": 1,
"id": "b1111111-1111-4111-8111-111111111111",
"name": "Opslagpool bijna vol",
"enabled": true,
"severity": "critical",
"scope": {
"entityType": "storage_pool"
},
"condition": {
"inputType": "metric",
"metric": "storage.pool.utilization.maximum",
"operator": ">=",
"threshold": 97,
"recoveryThreshold": 90,
"aggregation": "max",
"windowSeconds": 300
},
"evaluationIntervalSeconds": 60,
"pendingSeconds": 300,
"resolveSeconds": 300,
"unknownBehavior": "retain-firing-as-unknown",
"groupBy": [
"server"
],
"suppressWhen": [
"host.unreachable",
"storage.source.unavailable"
],
"message": {
"titleKey": "alerts.storagePoolCritical.title",
"bodyKey": "alerts.storagePoolCritical.body"
}
}
]
}
@@ -0,0 +1,55 @@
package alertdefaults
import (
"context"
"os"
"testing"
"time"
"github.com/itworx/pulse/internal/alert"
"github.com/itworx/pulse/internal/database"
"github.com/itworx/pulse/internal/metriccatalog"
)
func TestPostgreSQLDefaultSeedIsIdempotent(t *testing.T) {
dsn := os.Getenv("PULSE_TEST_DATABASE_URL")
if dsn == "" {
t.Skip("PULSE_TEST_DATABASE_URL is not set")
}
ctx, cancel := context.WithTimeout(context.Background(), 60*time.Second)
defer cancel()
pool, err := database.NewPool(ctx, database.Config{URL: dsn, MaxConns: 6, MinConns: 1})
if err != nil {
t.Fatal(err)
}
defer pool.Close()
if err := database.Migrate(ctx, pool); err != nil {
t.Fatal(err)
}
registry, err := metriccatalog.DefaultRegistry()
if err != nil {
t.Fatal(err)
}
repository := alert.Repository{Pool: pool, Registry: registry}
first, err := Seed(ctx, repository, registry, "system-defaults")
if err != nil {
t.Fatal(err)
}
second, err := Seed(ctx, repository, registry, "system-defaults")
if err != nil {
t.Fatal(err)
}
if second.Added != 0 || second.Existing != first.Added+first.Existing {
t.Fatalf("seed reports first=%+v second=%+v", first, second)
}
rules, err := Load(registry)
if err != nil {
t.Fatal(err)
}
for _, document := range rules {
loaded, err := repository.Get(ctx, document.ID)
if err != nil || loaded.ID != document.ID {
t.Fatalf("default %s was not readable after restart-safe seed: rule=%+v err=%v", document.ID, loaded, err)
}
}
}
+167
View File
@@ -0,0 +1,167 @@
package alertdefaults
import (
"context"
"testing"
"time"
"github.com/itworx/pulse/internal/alert"
"github.com/itworx/pulse/internal/metriccatalog"
)
type fakeStore struct {
rules map[string]alert.Rule
adds int
}
func (store *fakeStore) Create(_ context.Context, _ string, document alert.Document, _ string) (alert.Rule, alert.Version, error) {
if _, exists := store.rules[document.ID]; exists {
return alert.Rule{}, alert.Version{}, alert.ErrConflict
}
store.adds++
rule := alert.Rule{Document: document}
store.rules[document.ID] = rule
return rule, alert.Version{RuleID: document.ID, VersionNumber: 1}, nil
}
func (store *fakeStore) Update(_ context.Context, id, _ string, expected int64, document alert.Document, _ string) (alert.Rule, error) {
rule, exists := store.rules[id]
if !exists {
return alert.Rule{}, alert.ErrNotFound
}
if rule.Revision != expected {
return alert.Rule{}, alert.ErrConflict
}
rule.Document = document
rule.Revision++
store.rules[id] = rule
return rule, nil
}
func (store *fakeStore) Get(_ context.Context, id string) (alert.Rule, error) {
rule, exists := store.rules[id]
if !exists {
return alert.Rule{}, alert.ErrNotFound
}
return rule, nil
}
func TestLoadValidatesImplementationOwnedDefaults(t *testing.T) {
registry, err := metriccatalog.DefaultRegistry()
if err != nil {
t.Fatal(err)
}
rules, err := Load(registry)
if err != nil {
t.Fatal(err)
}
if len(rules) < 4 {
t.Fatalf("default rule count=%d, want at least 4", len(rules))
}
}
func TestSeedIsIdempotentAndDoesNotOverwriteExistingRule(t *testing.T) {
registry, err := metriccatalog.DefaultRegistry()
if err != nil {
t.Fatal(err)
}
rules, err := Load(registry)
if err != nil {
t.Fatal(err)
}
custom := rules[0]
custom.Name = "Aangepaste naam"
store := &fakeStore{rules: map[string]alert.Rule{custom.ID: {Document: custom}}}
first, err := Seed(context.Background(), store, registry, "system-defaults")
if err != nil {
t.Fatal(err)
}
if first.Added != len(rules)-1 || first.Existing != 1 || store.adds != len(rules)-1 {
t.Fatalf("first seed report=%+v adds=%d", first, store.adds)
}
second, err := Seed(context.Background(), store, registry, "system-defaults")
if err != nil {
t.Fatal(err)
}
if second.Added != 0 || second.Existing != len(rules) {
t.Fatalf("second seed report=%+v", second)
}
loaded, err := store.Get(context.Background(), custom.ID)
if err != nil || loaded.Name != "Aangepaste naam" {
t.Fatalf("custom rule was overwritten: rule=%+v err=%v", loaded, err)
}
}
func TestDefaultRulesReduceStormNoise(t *testing.T) {
registry, err := metriccatalog.DefaultRegistry()
if err != nil {
t.Fatal(err)
}
rules, err := Load(registry)
if err != nil {
t.Fatal(err)
}
var serviceRule alert.Document
for _, rule := range rules {
if rule.Name == "Service is niet bereikbaar" {
serviceRule = rule
}
}
if serviceRule.ID == "" {
t.Fatal("service default rule is missing")
}
at := time.Date(2026, time.January, 4, 12, 0, 0, 0, time.UTC)
signals := make([]alert.Signal, 0, 5)
for i := 0; i < 5; i++ {
signals = append(signals, alert.Signal{InstanceID: "service-" + string(rune('a'+i)), RuleID: serviceRule.ID, RuleVersionID: "version-1", EntityID: "entity-" + string(rune('a'+i)), Severity: serviceRule.Severity, State: alert.StateFiring, Fingerprint: "fingerprint-" + string(rune('a'+i)), EvaluationKey: "evaluation-1", ObservedAt: at, Labels: map[string]string{"host": "pulse", "application": "media", "service": "svc-" + string(rune('a'+i))}, GroupBy: serviceRule.GroupBy, SuppressWhen: serviceRule.SuppressWhen})
}
deduplicated, err := alert.DeduplicateSignals(append(signals, signals[0]))
if err != nil || len(deduplicated) != len(signals) {
t.Fatalf("deduplicated=%d err=%v", len(deduplicated), err)
}
groups, err := alert.GroupSignals(deduplicated)
if err != nil || len(groups) != len(signals) {
t.Fatalf("groups=%+v err=%v", groups, err)
}
decision, err := alert.EvaluateSuppression(signals[0], []alert.Cause{{Key: "dns.unavailable", State: alert.StateFiring, Confirmed: true, ObservedAt: at}})
if err != nil || !decision.Suppressed || decision.CauseKey != "dns.unavailable" {
t.Fatalf("suppression=%+v err=%v", decision, err)
}
}
func TestSeedRefreshesUntouchedDefaultButKeepsOperatorEdits(t *testing.T) {
registry, err := metriccatalog.DefaultRegistry()
if err != nil {
t.Fatal(err)
}
rules, err := Load(registry)
if err != nil {
t.Fatal(err)
}
stale := rules[0]
stale.PendingSeconds = stale.PendingSeconds + 60
edited := rules[1]
edited.Name = "Door operator aangepast"
store := &fakeStore{rules: map[string]alert.Rule{
stale.ID: {Document: stale, Revision: 1}, // seeded, never edited -> refreshed
edited.ID: {Document: edited, Revision: 2}, // edited by an operator -> untouched
}}
report, err := Seed(context.Background(), store, registry, "system-defaults")
if err != nil {
t.Fatal(err)
}
if report.Refreshed != 1 || report.Existing != 1 || report.Added != len(rules)-2 {
t.Fatalf("report=%+v", report)
}
refreshed, _ := store.Get(context.Background(), stale.ID)
if refreshed.PendingSeconds != rules[0].PendingSeconds || refreshed.Revision != 2 {
t.Fatalf("stale default not refreshed: %+v", refreshed)
}
kept, _ := store.Get(context.Background(), edited.ID)
if kept.Name != "Door operator aangepast" || kept.Revision != 2 {
t.Fatalf("operator edit overwritten: %+v", kept)
}
again, err := Seed(context.Background(), store, registry, "system-defaults")
if err != nil || again.Refreshed != 0 || again.Added != 0 {
t.Fatalf("second seed not idempotent: %+v err=%v", again, err)
}
}