Solrcloud 4.1 inconsistent # of results in replicas
I have a one shard collection, with one replica.
I did a dataImport from my oracle DB.
In the master, I have 93835 docs, in the non master 92627.
I have tried http://{machinename}:8080/solr/{collection}/update/commit=true
on the master, but the index does not replicate.
Also, the node list different generations of the core.
How do I get these nodes to sync?
Thanks
-Joey
--
View this message in context:
http://lucene.472066.n3.nabble.com/Solrcloud-4-1-inconsistent-of-results-in-replicas-tp4035638.html
Sent from the Solr - User mailing list archive at Nabble.com.
Re: Solrcloud 4.1 inconsistent # of results in replicas
Mark Miller-3 wrote > Does the admin cloud UI show all of the nodes as green? (active) > > If so, something is not right. > > - Mark Yes, the leader with the correct # is filled in green, and the other node is a green circle. -Joey -- View this message in context: http://lucene.472066.n3.nabble.com/Solrcloud-4-1-inconsistent-of-results-in-replicas-tp4035638p4035654.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: Solrcloud 4.1 inconsistent # of results in replicas
What can I provide to get more insight into this. I have tried lowering the commit maxDocs but the difference in nodes is several times the maxDocs. If I bring up a new node it will get the correct version, however if I kill the leader, the wrong version becomes the master, and replicates out. -Joey -- View this message in context: http://lucene.472066.n3.nabble.com/Solrcloud-4-1-inconsistent-of-results-in-replicas-tp4035638p4035739.html Sent from the Solr - User mailing list archive at Nabble.com.
Re: Solrcloud 4.1 inconsistent # of results in replicas
This is now JIRA issue SOLR-4343 -- View this message in context: http://lucene.472066.n3.nabble.com/Solrcloud-4-1-inconsistent-of-results-in-replicas-tp4035638p4035825.html Sent from the Solr - User mailing list archive at Nabble.com.
100% CPU usage in empty indexs
I have a puppet manifest that builds solr 4.1 inside of tomcat 7 & java 1.7.15. All goes well, all of the core initialize and so on, But the second I go to the admin page (host:8080/solr) the cpu usage on one core spikes to 100% and never comes back down. Nothing at all in the logs except for the warning that solr created a new index folder on startup. I'm not sure what is causing this. The JVM has a max heap of 4G and it's not even using 1, so I don't think it is GC related. Thanks -Joey -- View this message in context: http://lucene.472066.n3.nabble.com/100-CPU-usage-in-empty-indexs-tp4044292.html Sent from the Solr - User mailing list archive at Nabble.com.
