Am using external zookeeper ensemble (3.4.6) and solr cloud (5.3). Have 2 shards for a collection in solr cloud. Have 2 datacenters DC1 and Dc2. Each shard has 2 nodes, 1 on DC1 and another on DC2. We have added DC1 nodes to solr-DC1 load balancer and DC2 nodes to solr-DC2 load balancer. The application which sends the requests to solr cloud is on DC1.
1. When there is a search query from the application, will it hit first DC1 nodes in solr cloud ? 2. Only the leader does the write requests and then the replica replicates to itself. If the leader is a node on DC2, will it still first hit DC1 node and then go to DC2 node and writes there in DC2 node in the shard? 3. I have stopped DC1 nodes and still all writes and search requests work fine with DC2 nodes in the shard. 4. Want to know, if the 2 load balancers we created solr-DC1 and solr-DC2 makes any sense here ? 5. when a leader goes down in solr cloud shard, Does solr cloud itself re electes the leader and update in state.json in ZK znodes ? Or ZK elects the leader on solr cloud ? 6. How does a search query work in this case ? I have seen log entries on all of the 4 nodes for a search request. 7. I believe solr writes/search requests does not go through ZK. Please assist. Thanks, Ilango -- Sent from: http://lucene.472066.n3.nabble.com/Solr-User-f472068.html