Ishan Chattopadhyaya created SOLR-11816:
-------------------------------------------
Summary: Rolling restarts (rf>1) cause latency spike
Key: SOLR-11816
URL: https://issues.apache.org/jira/browse/SOLR-11816
Project: Solr
Issue Type: Bug
Security Level: Public (Default Security Level. Issues are Public)
Affects Versions: 6.6.2, 7.2
Reporter: Ishan Chattopadhyaya
During rolling restarts, even when replication factor >=2 (maxShardsPerNode=1),
there's a latency spike.
I've seen many of my clients being hit by this issue. In some cases during
rolling restarts, the shards get into a leaderless state (which I'm still
reproducing).
I've put together a reproduction suite using docker, influxdb and grafana.
https://github.com/chatman/solr-grafana-docker/tree/rolling-restart-test
Steps:
# Run {{docker-compose up}}
# Open Grafana at http://localhost:3000 (user: admin/pw: admin)
# Run indexing.sh and querying.sh both in separate terminals
# Let the graphs build up a bit, and then run rolling-restarts.sh
# A latency spike (about 4x) is observed.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
