[helpcodeai]

Summary

On 13 Oct at 12:16 UTC the Cloud backend died with FATAL ERROR: Zone Allocation failed - process out of memory. The heap was at about 1.4 GB of its 4.2 GB cap and the container was far from its 6 GB limit, so this was not the usual heap exhaustion, and the log had no stack to say what allocated it. Docker restarted it in about 35 seconds.

This makes the next fatal V8 crash leave a Node diagnostic report, so we can see what was running.

Changes

• start.sh: the backend command gets --report-on-fatalerror --report-exclude-env --report-compact --report-directory=$HEAP_SNAPSHOT_DIR (/app/diagnostics in the image, already a volume on Cloud for heap snapshots).
• The flags are added only when node --report-exclude-env works (Node >= 22.13). The report would otherwise contain every environment variable, secrets included, and an older Node would refuse to start on the unknown flag.
• docs/operations/observability.md: new section on the crash reports.

Cloud and self-hosted behave the same: same image, same script. Self-hosters keep the reports across container recreation by mounting a volume at /app/diagnostics.

Type

• Bug fix
• New feature
• Documentation
• Refactoring
• Performance improvement
• Other (describe below)

Testing

• Added/updated unit tests

• Tested manually (describe steps below)

• No tests needed (explain why)

• sh -n start.sh: OK.

• Extracted the flag logic and printed the resulting command: with Node 25 locally and HEAP_SNAPSHOT_DIR=/app/diagnostics it yields the flags above.

• Forced a heap OOM with the same flags locally: a compact JSON report was written, without environmentVariables.

• In the running Cloud container: Node v26.11.1 accepts --report-exclude-env, and /app/diagnostics is writable by appuser.

Checklist

• My code follows the existing code style
• I have tested my changes locally
• I have updated documentation (if applicable)
• All existing tests pass (npm test)
• This PR has a descriptive title

Related Issues

None.