Skip to content

chore(monitoring): raise Prometheus retentionSize 50GB->60GB - #95

Merged
hakan-persson merged 1 commit into
mainfrom
chore/prometheus-retention-size
Jul 28, 2026
Merged

chore(monitoring): raise Prometheus retentionSize 50GB->60GB#95
hakan-persson merged 1 commit into
mainfrom
chore/prometheus-retention-size

Conversation

@hakan-persson

Copy link
Copy Markdown
Member

What

Raises Prometheus retentionSize from 50GB to 60GB in the monitoring Helm values.

Why

retention: 90d and retentionSize: 50GB were both set; Prometheus enforces whichever is hit first. The 50GB size cap was the binding limit — the TSDB only held ~61 days of data instead of 90, i.e. it was already pruning the oldest data early. During camp week (higher ingest + new Traefik metrics) that deletion front would keep advancing toward the present, risking loss of early-camp data needed for the evaluation.

Raising the cap to 60GB (PVC is ~46GB used of ~67GB usable) makes 90d time retention the real limit again, keeping ~7GB raw-disk headroom for WAL/head/compaction.

Safety / rollout

  • No pod restart, no disk resize — just a Prometheus config value (live reload).
  • Deliberately does not change the 64Gi PVC request. A proper PVC grow is deferred: Prometheus currently runs on a Standard_A4m_v2 node that needs the disk detached to resize; the plan is to move it to the D8as_v5 userpool (soft affinity flip) and expand online later.

Follow-up

  • The Grafana PVC disk usage alert threshold needs adjusting — the Prometheus PVC will now legitimately sit near 60/67GB (~90%).

🤖 Generated with Claude Code

The 50GB size cap was the binding retention limit (below retention:
90d), so Prometheus was pruning the oldest data early — TSDB only held
~61 days instead of 90. During camp week the higher ingest would keep
advancing that deletion front toward the present, risking loss of
early-camp data needed for evaluation.

Raise the cap to 60GB (PVC is ~46GB used of ~67GB usable) so time-based
90d retention becomes the real limit, keeping ~7GB raw-disk headroom for
WAL/head/compaction. No pod restart or disk resize needed. A proper PVC
grow (and moving Prometheus to the D8as_v5 pool for online resize) is a
separate follow-up.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Copilot AI review requested due to automatic review settings July 28, 2026 07:38

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

This PR adjusts the kube-prometheus-stack configuration in this repo’s monitoring Helm values to increase Prometheus TSDB size-based retention, so the configured 90d time retention is less likely to be cut short by the size cap.

Changes:

  • Increased Prometheus retentionSize from 50GB to 60GB.
  • Added inline documentation explaining the rationale and PVC headroom considerations.

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

@hakan-persson
hakan-persson merged commit c4f0438 into main Jul 28, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants