Open kkalavantavanich opened 3 years ago
We have been plagued by this as well. No changes other than swapping from Kafka's Java SDK to alpakka-kafka.
There's likely an exception that's being swallowed. This is usually a result of the materialized value of the stream not being handled correctly.
In OP's case they should handle the Future[Done]
in DrainingControl.isShutdown
. This will resolve when the stream shuts down due to an error within the stream, or from without when shutting down the stream using the draining control itself (DrainingControl.shutdown
).
Thanks @seglo
Versions used
Scala version: 2.12.11 Akka version: 2.6.3 Alpakka-Kafka version: 2.0.2
Expected Behavior
Please describe the expected behavior of the issue, starting from the first action.
Actual Behavior
Please provide a description of what actually happens, working from the same starting point.
Relevant logs
Application logs was "committing xxx" (normal behavior) then it stopped coming without any errors. Application was deployed in docker and it didn't stop. Syslogs looks normal. Kafka audit logs looks normal. Other consumers still working and able to consume from Kafka. Lag for this consumer group increased.
Reproducible Test Case
My understanding of this bug is minimal. Bug is not reproducible. Please help provide provide possible ways to reproduce the bug.
Relevant code