Description
Build
4.5.0-1720
Perf sanity test:
./testrunner -i INI_FILE.ini -t fts.stable_topology_fts.StableTopFTS.test_query_type,items=1000000,num_queries=100,query_types=match,match_phrase,dataset=wiki,cluster=D+F,F,index_retry=20,get-cbcollect-info=True,query_max_matches=10000
The test tries indexing 1M wiki docs and querying it. Cluster: D+F:F (2 node cluster, 2 nodes with fts)
Indexing is stuck at 80.56% for more than 5 mins. Test enters query phase where all queries time out.
Test log : https://gist.github.com/arunapiravi/8e0ae6edd63de5356dfc
Live cluster : http://172.23.123.38:8091/ui/index.html
This could be related to moss but I need the cluster back to verify the no-moss case.
Attaching logs from both nodes.
Attachments
Issue Links
- is duplicated by
-
MB-18624 FTS: creating single index causes nearly 100% CPU on a small system and makes it hard to use
- Closed