openclaw - 💡(How to fix) Fix CLI backend responses sometimes not delivered to Telegram delivery context [1 comments, 2 participants]

• Session processes user message, generates assistant response via CLI backend
• Response appears in session transcript with provider: "claude-cli"
• No corresponding delivery-mirror message is created
• User never receives the response in Telegram
• Pattern is intermittent — some responses in the same session deliver fine

TL;DR

Implement a retry mechanism for failed deliveries in the delivery-mirror pathway to ensure assistant responses are delivered to Telegram.

Guidance

• Investigate the SessionWriteLockTimeoutError and pi-trajectory-flush timeout errors to understand the root cause of the lock contention and cleanup timeouts.
• Verify that the delivery-mirror pathway is correctly configured and enabled for sessions with a deliveryContext pointing to a Telegram chat/topic.
• Consider increasing the timeout values for SessionWriteLockTimeoutError and pi-trajectory-flush to reduce the likelihood of timeouts and subsequent silent delivery failures.
• Review the launchd watchdog workaround and consider integrating a similar mechanism to remove stale locks and prevent lock contention cascades.

Example

No code snippet is provided as the issue does not contain sufficient information to create a specific example.

Notes

The provided workaround only addresses the lock contention issue and does not solve the underlying problem of silent delivery failures. A more comprehensive solution is needed to ensure reliable delivery of assistant responses.

Recommendation

Apply a workaround by implementing a retry mechanism for failed deliveries in the delivery-mirror pathway, as the root cause of the issue is not fully understood and a comprehensive solution may require further investigation.