feat(server): get kubernetes resource by informer #213
Conversation
|
@codex review |
![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: e20519564f
ℹ️ About Codex in GitHub
Your team has set up Codex to 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 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| cached = informer.get(sandbox_id) | ||
| if cached: | ||
| return cached |
There was a problem hiding this comment.
This returns cached data before checking informer sync state, so after any watch/list failure (where _has_synced is flipped to false in WorkloadInformer._run) get_workload can still serve stale objects and never fall back to a direct API read for that sandbox ID. In practice, a sandbox deleted or changed while the informer is degraded can still be reported as existing here, which breaks read-after-delete and status accuracy for the agent-sandbox path.
Useful? React with 👍 / 👎.
| cached = informer.get(sandbox_id) | ||
| if cached: | ||
| return cached |
There was a problem hiding this comment.
The batchsandbox provider has the same stale-read path: cached objects are returned unconditionally before consulting has_synced, so once the informer loses sync it may continue returning outdated entries for existing cache keys and skip the API fallback. This can surface deleted BatchSandbox resources as still present until a successful resync, causing incorrect 200/404 behavior in production reads.
Useful? React with 👍 / 👎.
4 participants
Summary
Testing
Breaking Changes
Checklist