Debezium Server restarts after Pub/Sub warning: “attempted to publish batch with zero messages” Hi folks,
Hoang Vuong Xuan
I’m running two Debezium Server (3.1.2) instances that stream SQL Server CDC into Google Pub/Sub. Over the past few days, one instance has been restarting roughly once per day due to repeated TimeoutExceptions.
From the logs:
-
Error: Timeout waiting for publish results but seems like the root cause is debezium attempted to publish zero batch messages. I have no idea why this happens
What I’ve done so far:
-
Confirmed that the source connector is healthy (streaming ~35–90 records/sec, with occasional spikes).
- Increase debezium.sink.pubsub.wait.message.delivery.timeout.ms to 60000 ms
- Adjust Pub/Sub retry settings and possibly enable flow control
Attachments:
-
Logs from the affected instance
-
application.properties generated by Debezium Operator
Ask:
-
Has anyone seen “publish batch with zero messages” lead to timeouts/restarts?
-
Any known issues or config tweaks for Debezium Server 3.1.x + Pub/Sub that could cause zero-size batches?
-
Recommended diagnostics or additional metrics to capture?
Happy to share more details. Thanks in advance for your help!
Chris Cranford
It looks like this discussion was started and continued on Zulip [1], so if anyone is interested, let's continue the discussion there.
Thanks,
-cc
[1]: #community-dbz-server > Debezium Server: Attempted to publish zero messages batch @ 💬
--
You received this message because you are subscribed to the Google Groups "debezium" group.
To unsubscribe from this group and stop receiving emails from it, send an email to debezium+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/debezium/a86f742a-7f7d-47a9-ba6b-6664ff6818a8n%40googlegroups.com.