Microshardding question

[Jamie Johnson <je...@gmail.com>]

I have recently been considering using microsharding in our
distribution and wanted to make sure I understand the premise
correctly.  The idea would be to fire up a set of solr instances, each
with a number of cores.  Should those cores become over burdened we'd
fire up another solr instance on another box, add it to this shard in
the collection and solr would replicate the index from the leader.
Now if I wanted to remove the original over burdened core how would I
go ahead and update the clusterstate in zookeeper to no longer
reference the core that I am taking down?  I know that I may not
necessarily need to take it down, but I think there are cases where it
makes sense to do so.  I am assuming having it listed in clusterstate
isn't really a problem since the shard can be accessed from one of the
other solr instances but it would seem a good practice to remove this.
 Does deleting a solr core when running solrcloud remove the entry
from clusterstate?

Re: Microshardding question

[Jamie Johnson <je...@gmail.com>]

Seems like I should have just tried it.  Unloading did indeed remove
the core from the clusterstate, so that answers that.

On Sat, Mar 24, 2012 at 10:02 PM, Jamie Johnson <je...@gmail.com> wrote:
> I have recently been considering using microsharding in our
> distribution and wanted to make sure I understand the premise
> correctly.  The idea would be to fire up a set of solr instances, each
> with a number of cores.  Should those cores become over burdened we'd
> fire up another solr instance on another box, add it to this shard in
> the collection and solr would replicate the index from the leader.
> Now if I wanted to remove the original over burdened core how would I
> go ahead and update the clusterstate in zookeeper to no longer
> reference the core that I am taking down?  I know that I may not
> necessarily need to take it down, but I think there are cases where it
> makes sense to do so.  I am assuming having it listed in clusterstate
> isn't really a problem since the shard can be accessed from one of the
> other solr instances but it would seem a good practice to remove this.
>  Does deleting a solr core when running solrcloud remove the entry
> from clusterstate?