Remove readiness probe in favor of startup probe - #185
Conversation
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 256f136609
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| path: /health/readyz | ||
| port: rest | ||
| failureThreshold: 12 | ||
| periodSeconds: 5 |
There was a problem hiding this comment.
When /health/readyz takes longer than 60 seconds to return success (for example during a slow metastore/object-store dependency recovery or a cold cluster restart), this new startupProbe kills the container after 12 failures at 5s intervals; the old readinessProbe would simply keep the pod out of endpoints until the dependency recovered. Since the same readyz startupProbe defaults are repeated for the other Quickwit components, a transient >60s readiness delay can turn upgrades into CrashLoopBackOff instead of delayed readiness; please give the startup probe a much larger failureThreshold/period or keep readiness semantics separate.
Useful? React with 👍 / 👎.
1 participant
We decided to remove the readiness probe in quickwit-oss/quickwit#6673 and use a startup probe instead. In that PR, the startup endpoint is aliased to the readiness endpoint for backwards compatibility. This removes the readiness probe itself, and modifies the startup probe.