chore(monitoring): disable metric and log retention pruning

[hakan-persson]

Keeps all metrics and logs until the camp data has been exported. The cluster is being stopped shortly and restarted later to pull data, and nothing may be pruned in the meantime.

Prometheus

retention: 90d was not the binding constraint — the size cap was:

• 6 size-based deletions in the last 30 days, 0 time-based
• oldest sample was only 65 days old, not 90
• camp data was actively being deleted from the oldest end

Both are now disabled (retention: 0d, retentionSize: "0"). retentionSize is quoted so it renders as a string; a bare 0 would be an int and the Prometheus CR expects a size string.

Loki

Already kept data forever — compactor deletes are off by default and no retention_period was set anywhere in the live config. This pins both explicitly (retention_period: 0s, compactor.retention_enabled: false) so a chart-default change can't silently start pruning. No behaviour change.

Tradeoff

With no automatic pruning the Prometheus PVC no longer self-limits. Measured growth is ~0.78 GB/day against ~16.6 GB free — roughly 21 days of headroom, comfortable for the few days needed. Beyond that it becomes a hard wall, so the PVC needs watching rather than relying on a cap.

Note: the monitoring PVs are reclaimPolicy: Delete. That's fine for az aks stop/start (managed disks are preserved), but an ArgoCD prune of the monitoring app or a namespace deletion would destroy the data regardless of these settings.

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Disables automatic data pruning in the monitoring stack to ensure camp metrics and logs are retained until export, despite an upcoming cluster stop/start cycle.

Changes:

• Prometheus: disables both time- and size-based TSDB retention to prevent any pruning.
• Loki: explicitly pins infinite retention and disables compactor retention deletes to guard against future chart default changes.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
k8s/infra-manifest/monitoring/helm-values.yaml	Sets Prometheus retention and retentionSize to disable pruning, with updated operational notes.
k8s/infra-manifest/monitoring/loki-values.yaml	Explicitly sets Loki retention_period: 0s and compactor.retention_enabled: false to prevent retention from being enabled via defaults.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.