[ 
https://issues.apache.org/jira/browse/SOLR-3088?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13203741#comment-13203741
 ] 

Mark Miller commented on SOLR-3088:
-----------------------------------

My thoughts are in line with what Yonik has said.

I only saw this issue as:

dummy entries get added to the collection state - eg if you start the first 
node with -DnumShards=3, one real shard entry is added to the collection state 
and two dummy entries. As Yonik says, now all searches will fail because the 2 
dummy shards are not 'up' and the full collection is not available.

Again, as Yonik notes, a nice side effect is that you don't always have to pass 
numShards = except perhaps in the future when you want to reshard if that ends 
up making sense. Is it a little bit like storing the number of shards as you 
where? Heh - I guess it is - though with the side affect of nodes being able to 
see a full view of what shards should be in the collection right away (that is, 
the info is in the clusterstate.json).

I've been leaning towards getting the req to pass numShards every time out 
anyway having dealt with it for a while now. This seems like a slightly more 
useful way to do it compared to just storing that value though. As long as the 
system property would always update it, I never had much of a problem with 
storing the value actually. I had been imagining you would trigger a re shard 
by passing a new number - but I'm not even sure if that is how that will work 
out.

Essentially though, everything would work as it does now - I'm not sure we need 
to do anything in this issue?


                
> create shard placeholders
> -------------------------
>
>                 Key: SOLR-3088
>                 URL: https://issues.apache.org/jira/browse/SOLR-3088
>             Project: Solr
>          Issue Type: New Feature
>          Components: SolrCloud
>            Reporter: Yonik Seeley
>             Fix For: 4.0
>
>
> When creating a new collection, a placeholder for each shard should be 
> created.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: 
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

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

Reply via email to