Major Description
We have seen in several cases where the log cleaner either wasn't enabled or had experienced some failure that __consumer_offsets partitions can grow excessively. When one of these partitions changes leadership, the new coordinator must load the offset cache from the start of the log, which can take arbitrarily long depending on how large the partition has grown (we have seen cases where it took hours). Catching this problem is not always easy because the condition is rare and the symptom just tends to be a long period of inactivity in the consumer group which gradually gets worse over time. It may therefore be useful to have a broker metric for the load time so that it can be monitored and potentially alerted on. Same thing goes for the transaction log
Attachments
Issue Links
- links to
- mentioned in
-
Page Loading...