[NXP-30314] Kafka topic not available after creation - Nuxeo Issue Tracker

XML

Word

Printable

Details

Type: Bug
Status: Resolved
Priority: Minor
Resolution: Fixed
Affects Version/s: 10.10
Fix Version/s: 10.10-HF45, 11.x, 2021.2
Component/s: Streams

Tags:
- nxplatform
- platform-review
Sprint:
nxplatform #32
Story Points:
3

Description

When creating a topic in a Kafka cluster with replication the new topic may not be available immediately depending on the election/replication process.

This translates into "Unknown Log name" exception like:

2021-03-31T00:11:38,756 ERROR [main] [org.nuxeo.runtime.model.ComponentManager] Component service:org.nuxeo.runtime.stream.service notification of application started failed: Unknown Log name: holdDocumentsAction
java.lang.IllegalArgumentException: Unknown Log name: holdDocumentsAction
	at org.nuxeo.lib.stream.log.internals.AbstractLogManager.lambda$getAppender$3(AbstractLogManager.java:152) ~[nuxeo-stream-10.10-HF41.jar:?]
	at org.nuxeo.lib.stream.log.internals.AbstractLogManager.getAppender(AbstractLogManager.java:148) ~[nuxeo-stream-10.10-HF41.jar:?]
	at org.nuxeo.lib.stream.computation.log.LogStreamManager.lambda$initAppenders$1(LogStreamManager.java:161) ~[nuxeo-stream-10.10-HF41.jar:?]
	at org.nuxeo.lib.stream.computation.log.LogStreamManager.initAppenders(LogStreamManager.java:161) ~[nuxeo-stream-10.10-HF41.jar:?]
	at org.nuxeo.lib.stream.computation.log.LogStreamManager.register(LogStreamManager.java:73) ~[nuxeo-stream-10.10-HF41.jar:?]
	at org.nuxeo.runtime.stream.StreamServiceImpl.initProcessor(StreamServiceImpl.java:189) ~[nuxeo-runtime-stream-10.10-HF32.jar:?]
	at org.nuxeo.runtime.stream.StreamServiceImpl.start(StreamServiceImpl.java:160) ~[nuxeo-runtime-stream-10.10-HF32.jar:?]
	at org.nuxeo.runtime.model.impl.RegistrationInfoImpl.start(RegistrationInfoImpl.java:381) [nuxeo-runtime-10.10-HF10.jar:?]
	at org.nuxeo.runtime.model.impl.ComponentManagerImpl.startComponent(ComponentManagerImpl.java:709) [nuxeo-runtime-10.10-HF10.jar:?]

We need to wait for topic availability before continuing.
So far this has only been seen on 10.10 so it could be related to a Kafka bug in 2.1.1.