Handle intentional Granian SIGTERM shutdown

[harsh21234i]

Closes #7266 by avoiding a misleading “Unexpected exit” error when Reflex and its Granian worker receive SIGTERM together.

The worker watcher treats SIGTERM as intentional only after Granian's supervisor has received its own interrupt signal. Worker-only SIGTERM exits retain Granian's normal failure handling and release the listener when no workers remain.

Testing

• tests/units/utils/test_exec.py: 33 passed, 3 skipped
• git diff --check passed

[cubic-dev-ai]

All reported issues were addressed across 3 files

<sub>Reply with feedback, questions, or to request a fix.

Turn on auto-fix | Re-trigger cubic</sub>

[greptile-apps]

Confidence Score: 5/5

[Medium risk] Changes how the development backend handles worker shutdown signals.

The PR appears safe to merge; the earlier socket-release issue is fixed.

What we checked:

• Requests wait without a worker: A worker-only SIGTERM no longer sets interrupt_by_parent. When no worker remains alive, the watcher closes the listener.

Summary

The watcher now treats a Granian worker’s SIGTERM exit as intentional only when the supervisor is also stopping.

• Fixes the earlier finding: stopping only the worker still releases the listening socket.
• Adds a test for worker-only SIGTERM and closes sockets after the shutdown tests.
• Adds a short release note. No new actionable issues were found.

_{Reviews (2) · Last reviewed commit: "Fix Granian worker SIGTERM socket cleanu..." · Reviewed by Greptile}

[codspeed]

Merging this PR will improve performance by 8.74%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 1 improved benchmark
✅ 149 untouched benchmarks
⏩ 18 skipped benchmarks¹

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	test_hydration_metadata[200]	2 ms	1.8 ms	+8.74%

Tip

Curious why performance improved? Comment @codspeedbot explain why performance improved on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing harsh21234i:fix/7266-granian-sigterm-shutdown (d96e563) with main (8239259)}

Footnotes

1. 18 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[FarhanAliRaza]

this does not seem to fix the issue.

[FarhanAliRaza]

Closing in favor of #7538. In a full reflex run, the worker exits 0 from its own signal handler, and the supervisor usually has not handled the signal yet, so the exitcode == -SIGTERM and interrupt_signal check never matched (the error still showed in 10 of 10 group-SIGTERM shutdowns). #7538 treats exit code 0 as a requested stop instead. Thanks for the work on this, @harsh21234i.