Real-time monitoring dashboard for Proxmox VE, PBS, Docker, Kubernetes, TrueNAS and vSphere. Self-hosted, with alerts for failed backups, full disks, unhealthy containers and offline nodes.
Go aialertsdashboarddockerdocker-monitoringgohomelabinfrastructure-monitoringkubernetesmonitoringmspproxmoxproxmox-backup-serverproxmox-mail-gatewayproxmox-veself-hostedtruenastypescriptvspherewebhooks
Preserve reviewed monitoring repairs with current upstream overviews
pulse-triage[bot]e02f949b09ac
diff --cc internal/unifiedresources/registry_test.goindex 608e291880,6a721e846a..e7490bef07--- a/internal/unifiedresources/registry_test.go+++ b/internal/unifiedresources/registry_test.go@@@ -6908,23 -6908,42 +6908,63 @@@ func TestAgentNodeScanIndexTracksAgents}}+func TestRegistryBulkMetadataInvalidatesAtRecordBoundary(t *testing.T) {+ rr := NewRegistry(nil)+ rr.IngestRecords(SourceAgent, []IngestRecord{{SourceID: "host", Resource: Resource{Type: ResourceTypeAgent, Name: "before", Status: StatusOnline, LastSeen: time.Now().UTC(), Agent: &AgentData{AgentID: "host", Hostname: "before.example.test"}}}})+ rr.List()+ previous := rr.Hosts()[0]+ id := rr.ingestRecord(SourceAgent, "host", Resource{Type: ResourceTypeAgent, Name: "after", Status: StatusOnline, LastSeen: time.Now().UTC(), Agent: &AgentData{AgentID: "host", Hostname: "after.example.test"}}, ResourceIdentity{}, false)+ // This is the real record boundary, before parent/rollup batch finalization.+ actual := assertRegistryMetadataOracle(t, rr)+ if actual[0].Canonical.Hostname != "after.example.test" || rr.Hosts()[0].Name() != "after" {+ t.Fatal("record mutation reused pre-batch metadata or views")+ }+ if previous.Name() != "before" {+ t.Fatal("record mutation rewrote an already returned view")+ }+ rr.retainSupersededCanonicalIDs(id, []string{"agent:prior-era"})+ actual = assertRegistryMetadataOracle(t, rr)+ if len(actual[0].Canonical.SupersededIDs) != 1 || actual[0].Canonical.SupersededIDs[0] != "agent:prior-era" {+ t.Fatal("superseded identity retention failed to invalidate bulk metadata")+ }+}+++ func TestEvaluateResourceHealthKeepsStaleTelemetryBehindAWinningAlert(t *testing.T) {+ now := time.Date(2026, 10, 5, 1, 0, 0, 0, time.UTC)+ silent := Resource{+ ID: "docker-host-1",+ Type: ResourceTypeAgent,+ Status: StatusOnline,+ LastSeen: now.Add(-70 * time.Minute),+ SourceStatus: map[DataSource]SourceStatus{+ SourceDocker: {Status: "stale", LastSeen: now.Add(-70 * time.Minute)},+ },+ }++ // The alert still explains the verdict first, and the silence follows it+ // so tables can stop presenting the host's last readings as live.+ health := EvaluateResourceHealth(silent, []ResourceHealthAlert{{ResourceID: "docker-host-1", Level: "critical", Type: "docker-host-offline"}}, now)+ if health.Verdict != HealthCritical || len(health.Reasons) != 2 {+ t.Fatalf("expected critical with two reasons, got %+v", health)+ }+ if health.Reasons[0].Code != "critical_alert" {+ t.Fatalf("expected the alert to stay the first reason, got %+v", health.Reasons)+ }+ if health.Reasons[1] != (ResourceHealthReason{Code: "telemetry_stale", Detail: "1h"}) {+ t.Fatalf("expected a trailing telemetry_stale reason with the age, got %+v", health.Reasons[1])+ }++ // Without an alert the stale verdict carries the single reason as before.+ health = EvaluateResourceHealth(silent, nil, now)+ if health.Verdict != HealthStale || len(health.Reasons) != 1 || health.Reasons[0].Code != "telemetry_stale" {+ t.Fatalf("expected a lone stale reason, got %+v", health)+ }++ // A fresh source adds nothing behind an alert.+ fresh := silent+ fresh.SourceStatus = map[DataSource]SourceStatus{SourceDocker: {Status: "online", LastSeen: now}}+ health = EvaluateResourceHealth(fresh, []ResourceHealthAlert{{ResourceID: "docker-host-1", Level: "warning", Type: "cpu"}}, now)+ if len(health.Reasons) != 1 {+ t.Fatalf("expected only the alert reason for a fresh source, got %+v", health.Reasons)+ }+ }