How do I disable distributed search feature when I have only one shard

[pramodEbay <pr...@ebay.com>]

Hi there,

We have a solr cloud set up with only one shard. There is one leader and 15
followers. So the data is replicated on 15 nodes. When we run a solr query,
only one node should handle the request and we do not need any distributed
search feature as all the nodes are exact copies of each other.

Under certain load scenarios, we are seeing SOLRJ api is adding
isShard=true&distrib=false&shard.url=A,B,C etc. to all the queries.  Is the
solr query waiting for responses from A, B and C before returning back to
the client. If that is true, it is unnecessary and causing problems for us
under heavy load.

The thing is, somehow, these parameters are automagically added during query
time. How do we disable this. The solrj query that we build programatically
does not add these three parameters. Is there some configuration we can turn
on, to tell solrj not to add these parameters to the solr request.

Thanks,
Pramod



--
View this message in context: http://lucene.472066.n3.nabble.com/How-do-I-disable-distributed-search-feature-when-I-have-only-one-shard-tp4148449.html
Sent from the Solr - User mailing list archive at Nabble.com.

Re: How do I disable distributed search feature when I have only one shard

[Erick Erickson <er...@gmail.com>]

Are you using CloudSolrServer in your SolrJ program?

No matter what, the &distrib=false should be keeping the
query from going to more than one shard....

So I'd check the logs and see if the suspect query appears in
more than one node.

FWIW,
Erick


On Mon, Jul 21, 2014 at 4:13 PM, pramodEbay <pr...@ebay.com> wrote:

> Hi there,
>
> We have a solr cloud set up with only one shard. There is one leader and 15
> followers. So the data is replicated on 15 nodes. When we run a solr query,
> only one node should handle the request and we do not need any distributed
> search feature as all the nodes are exact copies of each other.
>
> Under certain load scenarios, we are seeing SOLRJ api is adding
> isShard=true&distrib=false&shard.url=A,B,C etc. to all the queries.  Is the
> solr query waiting for responses from A, B and C before returning back to
> the client. If that is true, it is unnecessary and causing problems for us
> under heavy load.
>
> The thing is, somehow, these parameters are automagically added during
> query
> time. How do we disable this. The solrj query that we build programatically
> does not add these three parameters. Is there some configuration we can
> turn
> on, to tell solrj not to add these parameters to the solr request.
>
> Thanks,
> Pramod
>
>
>
> --
> View this message in context:
> http://lucene.472066.n3.nabble.com/How-do-I-disable-distributed-search-feature-when-I-have-only-one-shard-tp4148449.html
> Sent from the Solr - User mailing list archive at Nabble.com.
>