Skip to content

CHANGELOG.md

Latest commit

 

History

History
161 lines (116 loc) · 7.78 KB

CHANGELOG.md

File metadata and controls

161 lines (116 loc) · 7.78 KB
Raw

Changelog

All notable changes to this Helm repository are documented in this file.

The format is based on Keep a Changelog.

[common-services 2.0.3] — 2026-07-24

Scope

Version bump and Ops hardening of common-services 2.0.3 for Kubernetes 1.34 / 1.35 clusters:

Domain Components
Backup / DR Velero
Databases CloudNativePG
Data processing Flink Kubernetes Operator
Observability / logs Alloy, Loki
Removal Nebula Operator (complete)

Delivery via Argo CD: a single final Git state common-services 2.0.3, one sync to that state. No intermediate Helm steps, no RC, no multi-sync migration runner.

Version table (current → target)

Component Current chart Current app Target chart Target app
Velero 10.1.2 1.16.2 12.2.0 1.18.2
CloudNativePG 0.28.3 1.29.1 0.29.1 1.30.1
Flink Kubernetes Operator 1.14.0 1.14.0 1.15.0 1.15.0
Alloy 1.2.1 v1.10.1 1.13.0 v1.20.0
Loki grafana/loki 6.40.0 3.5.3 grafana-community/loki 18.13.7 3.7.8
Nebula Operator 1.8.6 — removed —

Flink repository updated to https://downloads.apache.org/flink/flink-kubernetes-operator-1.15.0.

Delivery decisions

  • A single final Git state 2.0.3, one Argo CD sync.
  • No Helm steps / RC / multi-sync runner for Velero or Loki.
  • Velero rationale: official documentation is structured version-by-version; with the final CRDs applied by crds-installer and the target Velero image / AWS plugin, a direct 1.16 → 1.18 cut is acceptable for Ops (no need to deploy 1.17 as an intermediate step in this repository).
  • Loki rationale: the Grafana Community guide describes a helm upgrade after adapting values, not a chain of twelve intermediate deployments. Local values are corrected so they are consistent with S3 + SimpleScalable before the cut.

Why the Loki Helm repository changes

  • The OSS Loki Helm chart moved to Grafana Community.
  • On the old repository https://grafana.github.io/helm-charts, Loki charts 7.x+ target GEL (Grafana Enterprise Logs) only.
  • Do not use grafana/loki 7.1.0 (or any 7.x series from the historical Grafana repository) for this OSS stack.
  • Official migration link: https://grafana.com/docs/loki/latest/setup/upgrade/upgrade-to-community/
  • Target: repository https://grafana-community.github.io/helm-charts, chart 18.13.7, Loki application 3.7.8.

Breaking / Loki values

Required corrections in charts/common-services/values.yaml:

Topic Before (problematic) After (2.0.3)
Repository grafana/loki grafana-community/loki
deploymentMode SimpleScalable explicit SimpleScalable (community default = Monolithic)
loki.storage.type filesystem (inconsistent with S3 schema) s3
Schema object_store: s3, TSDB v13, period 2024-04-01 unchanged (no schema rewrite)
Compactor delete_request_store: filesystem delete_request_store: s3
Monitoring chart 18 monitoring.rules.alerting monitoring.alerts (+ monitoring.rules without the obsolete key)
ServiceAccount generated / not pinned serviceAccount.name: loki (IRSA / pod identity stability)
Backend PVC community default (whenDeleted: Delete + auto-delete) enableStatefulSetAutoDeletePVC: true + whenDeleted/whenScaled: Retain

S3 bucketNames / region remain documented placeholders; secrets and IRSA stay out of Git (environment overlays).

Overlays updated for the new repository / Nebula removal: charts/common-services/override-values.yaml, values-qaibtest.yaml.

Backup Manager

  • Image aligned: radiantone/eoc-backup-manager:1.18.2.
  • Removed the legacy documentation UI configuration and associated environment variables (Huma documentation at /docs, /openapi.json, /openapi.yaml).
  • Added backupManager.env (passthrough) for runtime overrides (e.g. HTTP_READ_HEADER_TIMEOUT).
  • Ports, probes, and LISTEN_PORT aligned on backupManager.service.containerPort.

Velero

  • Chart 12.2.0 / app 1.18.2.
  • AWS plugin aligned: velero/velero-plugin-for-aws:v1.14.4 (compatible with Velero 1.18.2).
  • upgradeCRDs: false kept — CRDs are applied by crds-installer at the pinned final version.
  • Expected runtime smoke test after sync: backup + restore.

CloudNativePG / Flink

  • CNPG chart 0.29.1 / operator 1.30.1; CRDs via crds-installer; local values unchanged except for incompatibility (none required in this cut).
  • Flink Operator 1.15.0 + versioned Apache URL; CRDs via crds-installer; webhook remains create: false.

Alloy

  • Chart bump 1.13.0 / app v1.20.0 only.
  • DaemonSet topology kept (controller.type: daemonset).
  • No HA clustering / StatefulSet in this cut.

Nebula — complete removal

Ops assumption: no application still uses Nebula. If CRs remain in the cluster, they are deleted anyway (not fail-closed).

Removed from the chart:

  • nebula-operator dependency in Chart.yaml;
  • nebula-operator values block;
  • overlays (override-values.yaml, values-qaibtest.yaml);
  • workaround template templates/nebula/controller-manager-patch-rbac.yml;
  • Nebula dashboards (dashboards/ido/graph-database.json, dashboards-logs/ido/nebula-*.json);
  • backup-manager RBAC specific to Nebula;
  • Nebula install entry in crds-installer (no chart left to install).

crds-installer extension (no separate Job/hook):

  1. config entry action: delete / cleanup: true for the \.nebula-graph\.io pattern;
  2. list matching CRDs;
  3. for each CRD, purge all CR instances (cluster-wide / namespaced);
  4. remove blocking finalizers if needed;
  5. delete the CRDs;
  6. no-op if already absent;
  7. does not fail because CRs existed — they are purged (|| true on cleanup).

Job RBAC widened with verbs on apps.nebula-graph.io and autoscaling.nebula-graph.io.

Validation

Git render (required before merge):

helm dependency update charts/common-services
helm lint charts/common-services
helm template test charts/common-services --kube-version 1.34.0 >/dev/null
helm template test charts/common-services --kube-version 1.35.0 >/dev/null

Runtime checklist, non-prod then prod:

  • Manifest diff limited to the five upgraded components + Nebula removal
  • Velero: backup/restore smoke test after sync
  • Loki: push + query historical S3 logs + new writes; no prolonged mixed-version window
  • CNPG / Flink: operator Ready, existing CRs functionally unchanged
  • Alloy: pods Ready, logs to Loki gateway, Prometheus scrapes
  • Nebula: operator absent, CRs absent, CRDs absent, sync Healthy

Official links

Out of scope (intentional)

  • Velero 1.17 / Loki 6.55 intermediate steps and associated scripts/RCs
  • Alloy topology redesign (HA clustering)
  • Moving Loki off SimpleScalable (to be planned before Loki 4.0, not in this cut)
  • velero-ui bump unless blocked by the new Velero (not observed as blocking here)

This file is the Ops source of truth for the common-services 2.0.3 upgrade.