fix(collector): isolate and drain complete undrained LOB segments - #1156
fix(collector): isolate and drain complete undrained LOB segments#1156proerror77 wants to merge 2 commits into
Conversation
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Advanced Run ID: 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. Comment |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: f28ccd8cba
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".
f28ccd8 to
b803762
Compare
Recovery isolate previously skipped a spool with no incomplete parts, so rotated complete manifests sat on the live spool after archiver death. Isolate now queues those undrained complete segments; drain still uploads them with --upload-only and does not invent a resume identity. Co-authored-by: Wild Card <proerror77@users.noreply.github.com>
Do not isolate leftover uploaded-cleanup markers as undrained segments. Drain refuses the fresh-upload verify path when only those markers remain, so the marker is not deleted and the failed job stays inspectable. Co-authored-by: Wild Card <proerror77@users.noreply.github.com>
b803762 to
3ef8c03
Compare
Change
LOB recovery isolate now queues rotated complete undrained segments (
*.manifest.json,*.jsonl.zst,._SUCCESS), not only incomplete*.jsonl.part/*.zst.tmp. After archiver death, those sealed local segments had no drain path except the 600s/900s upload-staleness monitor.Leftover
*.uploaded-cleanup.jsonmarkers are not undrained segments. Isolate leaves them on the live spool for the production archiver. Drain of a cleanup-marker-only job fails closed without--upload-only/ fresh triplet verify, so the marker is not deleted and the failed job remains inspectable. Failed jobs still require an identity-bearingresume.Issue relationship
None
Validation
git diff --checkdeployment/aliyun/test-rust-lob-recovery-queue.sh— helpers reject empty and cleanup-marker-only spools, isolate queues complete undrained segments without queueing leftover cleanup markers, complete-only drain runs--upload-onlywithout--recover-parts-only, and cleanup-marker-only drain fails without entering fresh-upload verify.Runtime impact and rollback
Collector recovery only. After the next controller/runtime deploy of
host-rust-lob-recovery-queue.sh, productionExecStartPreisolate will move pending complete segments into the recovery queue and will not isolate leftover cleanup markers. Rollback is the previous recovery controller. This is not a live trading change and does not auto-resume failed jobs.