On 8/28/2013 1:36 PM, Jared Griffith wrote:
We are using Java here. Are you saying that the Solr java client would be
aware of the multiple zookeepers and would thus do health / host checks on
each zookeeper instance in turn until it got one that is working (assuming
that you have one or more zookeepers down)?
If that's the case, holy awesome.
I'll probably jump in IRC when I actually tackle this set up later on today.
Yes, the Java client is completely aware of the cloud state in realtime.
When you create a CloudSolrServer object, you don't tell it where Solr
is, you tell it where zookeeper is - using the same (potentially
multi-host and including a chroot) zkHost parameter that you give to Solr.
Thanks,
Shawn