Jason Gerlowski created SOLR-17941:
--------------------------------------
Summary: MOVEREPLICA should use a larger timeout (with a
configurable override)
Key: SOLR-17941
URL: https://issues.apache.org/jira/browse/SOLR-17941
Project: Solr
Issue Type: Improvement
Components: SolrCloud
Affects Versions: main (10.0)
Reporter: Jason Gerlowski
There are (arguably) two problems with how Solr's "move-replica" API handles
timeouts:
# we document a "timeout" parameter and the MoveReplicaCmd code appears to
support it, but the v1 API code in CollectionsHandler doesn't actually tunnel
that parameter through to the overseer.
# in the absence of a user-specified timeout, move-replica defaults to 10m
which (IMO) isn't quite adequate for today's index sizes (which tend to be much
larger than when the API was added back in 2017)
(2) is a bit of an opinion, so open to other perspectives there. But (1) is
pretty inarguably a bug. So this ticket should definitely fix (1), and
optionally fix (2) as well.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
