Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Not A Problem
-
None
-
None
-
None
-
Small test cluster (~20 nodes)
Description
I was looking at a client that was ingesting slowly with dlmarion. We jstacked the client to determine what was going on. We found 700+ BatchWriterLatencyTimer threads. This client only has one BatchWriter.
The same client was known to have zookeeper disconnects, and servers reported that sessionIDs could not be found. Suspect that the client was running out of memory, having GC pauses, which caused zookeeper disconnects and session timeouts.
The same client was running well under 1.6, and began having issues under 1.7.
Attachments
Issue Links
- relates to
-
ACCUMULO-4090 BatchWriter close not cleaning up all resources
- Resolved