[
https://issues.apache.org/jira/browse/SOLR-6816?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14248447#comment-14248447
]
Timothy Potter commented on SOLR-6816:
--------------------------------------
bq. Yep. Just like on a single node, if "overwrite=false", we could skip
version checking.
[~ysee...@gmail.com] I'm curious if we shouldn't have a different flag here,
something like "bulkAdd=true" to disable the version lookup to the ulog on the
replicas for bulk adds vs. trying to overload the overwrite=false behavior as
it seems like setting overwrite=false would by-pass a lot of checking around
the unique doc ID. This is how I think it should work:
* client sends batch of new docs with bulkAdd=true to leader
* leader does versionAdd
* leader sends docs with version set to replica(s) and bulkAdd=true param
* replica checks bulkAdd and by-passes the lookup to the ulog for that doc
I guess what I need to hear is that we think it's safe and a valid approach to
allow an indexing client to tell Solr this request contains a bunch of new
docs, so optimize for that. Thanks in advance for feedback ;-)
> Review SolrCloud Indexing Performance.
> --------------------------------------
>
> Key: SOLR-6816
> URL: https://issues.apache.org/jira/browse/SOLR-6816
> Project: Solr
> Issue Type: Task
> Components: SolrCloud
> Reporter: Mark Miller
> Priority: Critical
> Attachments: SolrBench.pdf
>
>
> We have never really focused on indexing performance, just correctness and
> low hanging fruit. We need to vet the performance and try to address any
> holes.
> Note: A common report is that adding any replication is very slow.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
