[Bug] T1K broken pipe occurring every 30-60 minutes

What happened?

Issue: T1K broken pipe occurring every 30-60 minutes
Version: 9.2.5 and 9.2.7 (persists after update)
Frequency: ~7 times in 4 hours
Pattern: Random, not correlated with traffic spikes or resource pressure
Memory: 101MB / 11GB used
CPU: 4.46%
Logs: "flush T1K packet error: Broken pipe (os error 32)"

How we reproduce?

1. ...
2. ...
3. ...

Expected behavior

No response

Error log

No response

Thanks for reporting this.

Broken pipe (os error 32) means that T1K attempted to write a detection packet after the connection to the detector had already been closed. This is a connection-level symptom, and the CPU/memory usage alone is not sufficient to determine the root cause.

Versions 9.2.5 and 9.2.7 are no longer current. Please first upgrade to the latest release (currently v9.3.10) and check whether the issue can still be reproduced.

If it still occurs, please provide:

1. The exact occurrence timestamp and timezone.
2. Sanitized logs from both safeline-tengine and safeline-detector, covering about one minute before and after the error.
3. Whether any request failed, returned a 5xx response, or experienced an unexpected detection/protection result at the same time.
4. The container status and restart counts:
   docker inspect -f '{{.Name}} status={{.State.Status}} restart_count={{.RestartCount}} started_at={{.State.StartedAt}}' safeline-tengine safeline-detector
5. Whether either container or the Docker network is restarted periodically.

Please redact any sensitive information from the logs before posting them.