The metrics reported by Kafka Workmanager after inactivity are invalid, reporting lots of scheduled job.
After some activity in all the partitions, the metrics return to normal.
When Nuxeo is restarted after a long inactivity, all messages are reprocessed.
This is because the consumer positions are not persisted after offsets.retention.minutes which is 1140 or 1 day.
See related Kafka tickets:
This is fixed by configuring Kafka to have an offsets retentions longer than the log retention.
- is related to
-
NXP-22599 MQueue kafka expose consumer lag
- Resolved
- links to