I have an ES cluster of 3 servers where indexes are configured with 5 shards and 2 replicas (so, every index has 5 primary shards and 10 replica shards, with 5 shards allocated to each server). I have just upgraded from 1.0.0 to 1.3.4 by stopping one server at a time, updating ES then re-starting it, waiting for all of the unallocated shards to initialize on the updated server.
Before performing the upgrade, the primary shards for each index were distributed evenly amongst the 3 ES servers, but now every primary shard is allocated to a single server. This appears to be causing an issue with snapshots as the server holding all of the primary shards is doing all the work to push a snapshot out to S3 while the other 2 servers are doing nothing - with even primary shard distribution each of the servers would be doing part of the work. Is there a way to ask ES to re-allocate the primary shards evenly amongst all the servers (like you get on a new index creation) so we can spread our snapshot load around? -- You received this message because you are subscribed to the Google Groups "elasticsearch" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/4f416b99-9656-400e-9de5-519bdac35a2c%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
