[Infra] Reduce llm_translation_testing parallelism and tolerate worker restarts - #25898
Conversation
…orker restarts Workers in llm_translation_testing have been crashing mid-run with "Not properly terminated" (OOM), even after bumping resource_class to xlarge. Reduce xdist workers from 8 to 4 to lower peak memory, and add --max-worker-restart=5 so a crashed worker is replaced instead of failing the whole run.
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Greptile SummaryThis PR reduces the Confidence Score: 5/5Safe to merge — minimal, targeted CI config change with no impact on application code or test logic. Single CI config file changed; only the pytest invocation for No files require special attention.
|
| Filename | Overview |
|---|---|
| .circleci/config.yml | Reduces llm_translation_testing xdist workers from -n 8 to -n 4 and adds --max-worker-restart=5 to tolerate OOM-induced worker crashes |
Flowchart
%%{init: {'theme': 'neutral'}}%%
flowchart TD
A[llm_translation_testing job starts\nresource_class: xlarge] --> B[Install deps\nuv 0.10.9 + checksum verified]
B --> C[pytest -n 4\n--timeout=120\n--max-worker-restart=5\n--retries 2]
C --> D{Worker crash?}
D -- No --> E[Test completes normally]
D -- Yes, restarts ≤ 5 --> F[Pending tests redistributed\nto remaining workers]
F --> D
D -- Yes, restarts > 5 --> G[Job fails]
E --> H[Store JUnit results]
Reviews (1): Last reviewed commit: "[Infra] CI: reduce llm_translation_testi..." | Re-trigger Greptile
…omMitigation_staging [Infra] Reduce llm_translation_testing parallelism and tolerate worker restarts
Relevant issues
Summary
Problem
Workers in the `llm_translation_testing` CircleCI job have been crashing mid-run with `[gw*] node down: Not properly terminated`, failing the job and blocking the pipeline. The crashes are OOM kills: multiple workers go down near the end of the run (97–99% through), and the log shows accumulating `Unclosed client session` / `Unclosed connector` messages before the final crash. Bumping `resource_class` from `large` to `xlarge` did not resolve it.
Fix
Two small, independent mitigations on the pytest invocation:
Staging counterpart of #25897.
Testing
CI will exercise the new configuration when this PR runs.
Type
🚄 Infrastructure