Details
-
Bug
-
Resolution: Duplicate
-
Critical
-
5.5.0
-
None
-
Untriaged
-
Ubuntu 64-bit
-
Unknown
Description
Noticed this on our perflab (http://172.23.121.90:8091/) cluster which is storage for all performance results and backend database for the ShowFast.
Looks like it happened after upgrade from 5.1 to 5.5 but I don't know for sure.
At some point some random indexes started growing in size until it take whole disk space and node starts throwing disk errors. Amount of documents didn't change.
cbcollect from one of the nodes:
scp root@172.23.104.212:/root/index_grows.zip index_grows.zip
I've also copied index folder there:
scp -r root@172.23.104.212:/root/mainIndex ..
There were two indexes "in trouble" on that node:
weekly_weekly_by_build
clusters_clusters_by_name
Here is what's taking space:
Attachments
Issue Links
- relates to
-
MB-30870 BACKPORT [plasma] logCleaner runs infinitely hanging warmup
- Closed