lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Erick Erickson (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (SOLR-9273) Share and reuse config set in a node
Date Fri, 01 Jul 2016 16:20:11 GMT

    [ https://issues.apache.org/jira/browse/SOLR-9273?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15359209#comment-15359209
] 

Erick Erickson commented on SOLR-9273:
--------------------------------------

I started looking at this along side shareSchema and quickly gave up due to what you've outlined.
Actually I didn't get that far, just poked around long enough to say "Oh my, this is complicated"...

So would this mean that there's one way of loading solrconfig in stand-alone and one for SolrCloud?
And, if so, would it make sense to deal with Zookeeper as "the one source of truth" first?
Or would we impose the same restrictions on stand-alone Solr re: point <1>? I'm not
against any of those options, just wondering what the thinking is here...

But this would be great for those situations where people have lots of replicas sharing a
configset. I did some timings a while ago for core discovery and it was something like 1,000/second
for just the discovery part. Might it be possible to hit 100 cores/second with this?

Just had a random thought here. Currently core loading implicitly throttles ZK state changes,
so I do wonder at the effects on Zookeeper/event notification to the clients... I can't point
to anything bad here, more of "gee, I wonder how that'll work"....

> Share and reuse config set in a node
> ------------------------------------
>
>                 Key: SOLR-9273
>                 URL: https://issues.apache.org/jira/browse/SOLR-9273
>             Project: Solr
>          Issue Type: Improvement
>      Security Level: Public(Default Security Level. Issues are Public) 
>          Components: config-api, Schema and Analysis, SolrCloud
>            Reporter: Shalin Shekhar Mangar
>             Fix For: 6.2, master (7.0)
>
>
> Currently, each core in a node ends up creating a completely new instance of ConfigSet
with its own schema, solrconfig and other properties. This is wasteful when you have a lot
of replicas in the same node with many of them referring to the same config set in Zookeeper.
> There are many issues that need to be addressed for this to work so this is a parent
issue to track the work.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org


Mime
View raw message