0

I'm trying to deploy Solr Cloud into Kubernetes using StatefulSets. My problem is that if I'm writing into Solr during a rolling update of the statefulset I get errors when I reach the deployment of the solr cluster leader. With just reads I can deploy Solr Cloud without any problems - using the SolrCloudClient in java. I've already tried rebalancing the leader on the preStop Hook of the leader but I still get errors on the SolrCloudClient writes... Does anyone have any tips on how to do a rolling update of Solr Cloud without any errors even during writes? Thank you

Bruno René Santos
  • 401
  • 4
  • 10
  • _I still get errors on the SolrCloudClient writes_ what is the specific error you receive? – mdaniel Mar 18 '18 at 21:17
  • Usually the errors are about the client not being able to connect to the previous leader. I have read this https://lucene.apache.org/solr/guide/6_6/read-and-write-side-fault-tolerance.html and it seems a partial solution is to separate the data into multiple shards... but I would like to know if there are any strategies that we can follow other than retrying the erroneous requests – Bruno René Santos Mar 18 '18 at 23:11

0 Answers0