[
https://issues.apache.org/jira/browse/SOLR-7344?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14509793#comment-14509793
]
Hrishikesh Gadre commented on SOLR-7344:
----------------------------------------
[~yo...@apache.org] Yes I think this can work. Although a minor point -
typically every endpoint is associated with a thread-pool and an accept queue.
Hence when all threads are busy, one or more outstanding requests can still
stay in the accept queue without being rejected. In our case, this facility
will probably be not available (since we would need to reject the request right
away to free up the thread). I am not saying this is a deal-breaker, but just
something I realized while thinking about this.
Anyone else has any thoughts on this? Or should I take a crack at this
alternate design?
> Use two thread pools, one for internal requests and one for external, to
> avoid distributed deadlock and decrease the number of threads that need to be
> created.
> ---------------------------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: SOLR-7344
> URL: https://issues.apache.org/jira/browse/SOLR-7344
> Project: Solr
> Issue Type: Improvement
> Components: SolrCloud
> Reporter: Mark Miller
> Attachments: SOLR-7344.patch
>
>
--
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
