[
https://issues.apache.org/jira/browse/SOLR-6820?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Timothy Potter updated SOLR-6820:
---------------------------------
Attachment: SOLR-6820.patch
First pass ... raising this to 65536 as Mark indicated made a significant
impact in overall indexing performance in my testing.
Specifically, with branch_5x my baseline got:
{code}
num docs secs docs/sec
9992262 427 23401.08 (no replication, leader only)
9992262 758 13182.40 (leader + 1 replica)
{code}
with this patch applied and using 65536:
{code}
9992262 382 26157.75
9992262 710 14073.61
{code}
that's a pretty significant bump in overall performance for an additional 0.5M
of memory per collection IMO
> The sync on the VersionInfo bucket in DistributedUpdateProcesser#addDocument
> appears to be a large bottleneck when using replication.
> -------------------------------------------------------------------------------------------------------------------------------------
>
> Key: SOLR-6820
> URL: https://issues.apache.org/jira/browse/SOLR-6820
> Project: Solr
> Issue Type: Sub-task
> Components: SolrCloud
> Reporter: Mark Miller
> Attachments: SOLR-6820.patch, threads.png
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
