[
https://issues.apache.org/jira/browse/SOLR-6816?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14384479#comment-14384479
]
Yonik Seeley commented on SOLR-6816:
------------------------------------
bq. For instance, the leader knows if a doc didn't exist (because it's doing a
similar ID lookup) [...]
I think the leader doesn't need to do this lookup since it is the one
establishing the order? Although there is other stuff like peer sync and
recovery that complicate the picture.
Anyway, the primary reordering issue is between the leader and replica, so even
if the leader says "hey, previous doc exists for id:1234", it could be
reordered and hence be false by the time it gets to the replica.
Longer term, I'd like to see reordering eliminated. The primary mechanism
would be via a single communication channel between the leader and it's
replicas that would enforce ordering.
> 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: [email protected]
For additional commands, e-mail: [email protected]
