Thank you for highlighting the actual issue that needs to be addressed.
At this point, I don't know if there is any best answer as to where in the
cluster aggregate operations like Data Import Handler and Solr Cell should
be run. My own inclination is that both of them need to be pulled out of
core Solr and run in a distributed manner outside of the core of a Solr
Cloud cluster.
DIH in particular is likely to produce a wide range of key values that will
distribute across the entire cluster, so picking a particular shard or node
to run it will not have any particular advantage.
This is an interesting issue to ponder.
-- Jack Krupansky
-----Original Message-----
From: kowish.adamosh
Sent: Saturday, July 13, 2013 10:51 AM
To: solr-user@lucene.apache.org
Subject: Re: SolrCloud leader
Ya :-) I need it for different reason.
I have 8 server nodes and one of them has to invoke scheduled data import
(by invoking CloudSolrServer)... And that's why I need to identify exclusive
server (it doesn't have to be leader but I think it is the simplest
solution). I don't want to invoke scheduled updates in separate application
because of HA requirement. Any ideas?
Kowish
--
View this message in context:
http://lucene.472066.n3.nabble.com/SolrCloud-leader-tp4077759p4077806.html
Sent from the Solr - User mailing list archive at Nabble.com.
