Open kaojunsong opened 1 year ago
I searched in the Slack, and found that someone suggest to use
"context": {
"forceTimeChunkLock": false
}
And the compaction task works after I added this setting. But I don't know if there is any side effect if I add this setting.
The indexSpec is not ready
warning happens if the task is unable to acquire a lock on its interval. The interval is 1970-01-01T00:00:00Z/2023-11-18T07:14:34.000Z
(from your compact
task spec) so it would be trying to get an exclusive lock for that interval. Is it possible some other task was running at the time? If so, that would cause the behavior you saw. You can retry later when no other tasks are running.
And the compaction task works after I added this setting. But I don't know if there is any side effect if I add this setting.
forceTimeChunkLock: false
is an experimental setting which you can read more about here: https://druid.apache.org/docs/latest/ingestion/tasks.html#locking. It makes the locking more fine-grained, so avoids the lock contention issue that you are probably hitting. It is experimental, though, so I'd only recommend it if you are comfortable with a feature that is not currently viewed as production-ready.
An update — I believe this is the same bug as described here: https://github.com/apache/druid/issues/13540#issuecomment-1344899192.
Hi experts,
I am using druid 0.22.1, and I want to compact some small segments, so I submit a compaction task:
But the compaction task failed with the following status information
And the logs:
I notice there is a "indexSpec is not ready:" warning in the log, but I don't know what does this mean. FYI: The datasource is "Fully available" Can anyone help me? Or how can I troubleshout issues like this? Thanks a lot in advance.