fix: Verify stream state before triggering VOD recording (#552)

[filthyrake]

Summary

• Re-fetches stream from the database after the ending -> ended status update to verify the state actually persisted before triggering VOD recording
• Prevents VODs from being created with incomplete or inconsistent stream data if the update was rolled back or lost
• Logs a warning when verification fails so the issue is observable

Closes #552

Test plan

• Verify normal VOD recording flow still works when stream correctly transitions to "ended"
• Confirm that VOD recording is skipped (with warning log) if stream status doesn't match after update
• Check that the fetch_one_with_retry re-fetch handles database contention gracefully

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Improves reliability of live stale-stream cleanup by verifying the stream’s persisted state after transitioning ending -> ended before triggering VOD recording, addressing Issue #552.

Changes:

• Adds fetch_one_with_retry usage to re-fetch the stream row after the ended update.
• Gates VOD finalization/recording on the verified post-update stream status.
• Logs a warning when post-update status verification fails.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The VOD trigger condition is now based on updated_stream["auto_record_vod"], which can change behavior if that flag is modified between the initial SELECT and the re-fetch. Elsewhere (e.g., end_live_stream) the decision to trigger VOD recording is based on the original row that drove the transition, using the re-fetched row only for verification/response. Consider keeping the verification limited to status == "ended" and using the original stream["auto_record_vod"] for the VOD decision to avoid unexpected skips/starts due to concurrent edits.

Suggested change

	if updated_stream and updated_stream["status"] == "ended" and updated_stream["auto_record_vod"]:
	try:
	await finalize_playlists_for_vod(stream["id"])
	await trigger_vod_recording(stream["id"])
	except Exception as e:
	logger.error(f"Failed to create VOD for stream {stream['slug']}: {e}")
	if updated_stream and updated_stream["status"] == "ended":
	if stream["auto_record_vod"]:
	try:
	await finalize_playlists_for_vod(stream["id"])
	await trigger_vod_recording(stream["id"])
	except Exception as e:
	logger.error(f"Failed to create VOD for stream {stream['slug']}: {e}")

[Copilot]

fetch_one_with_retry(...) can raise (e.g., DatabaseTimeoutError / DatabaseRetryableError). As written, a failure to re-fetch will abort detect_stale_streams() and prevent processing any remaining ending streams, even though the status update already succeeded. Consider wrapping the re-fetch + verification in a try/except and treating exceptions as a verification failure (log warning + skip VOD) so stale detection continues for other streams.