[rhanka]

The new monthly snapshot was created successfully, but production Elasticsearch refused to allocate its 15.3 GB restore shard because two prior 15 GB generations filled the 45 GiB PVC beyond the 90% disk watermark. The restore job now recovers from an incomplete red target, preserves the live alias target, removes only older unreferenced restore generations before a real restore, and retains all S3 snapshots.

The monthly wrapper timeout is raised from 10 to 125 minutes so it can wait for the child release-prod job, whose own timeout is 120 minutes.

Validation:

• production evidence: live alias resolved to the August index; only the incomplete target and July rollback are selected for deletion
• embedded restore script passes POSIX sh -n
• manifest parses and passes kubectl apply --dry-run=client
• prod overlay passes kubectl kustomize
• git diff --check
• AGY Gemini 3.8 high second-pass review: PASS, no blocker/high/medium findings

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: f9085856-7683-452e-9589-bf90fa9b5f6e

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}