[
https://issues.apache.org/jira/browse/SOLR-9847?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15761858#comment-15761858
]
Steve Rowe commented on SOLR-9847:
----------------------------------
I beasted {{TestBulkSchemaConcurrent}} for 500 iterations with and without the
patch, and the patch doesn't seem to change anything: unpatched, 27/500
iterations had the 600 second timeout; and patched, 28/500 iterations did.
[[email protected]], did your patch improve things for you locally?
A different idea: we could release the update lock after local changes are
made, but before waiting for all replicas to receive the update. I'll try this
out.
> Deadlock on ManagedIndexSchema lock.
> ------------------------------------
>
> Key: SOLR-9847
> URL: https://issues.apache.org/jira/browse/SOLR-9847
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Reporter: Mark Miller
> Attachments: SOLR-9847.patch
>
>
> Seems we hold the lock while in
> ManagedIndexSchema.waitForSchemaZkVersionAgreement, so we may never see
> agreement because are updates may also be waiting on that lock.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
