Since the upgrade to Elasticsearch 5.0 we are seeing a lot of movement of shards. In fact, at times there are node_failed exceptions received, but the node seems to be healthy and running elasticsearch correctly. There are spikes in indexing latencies seen at these moments. To investigate the same issue, I was trying to see the reason why Elasticsearch performed so many shard movements but couldn't find a relevant log message. The nodes data nodes are m4.2xlarge and masters are t2.medium.
I know Monitoring does show the history of shard allocation movements, but it doesn't tell the reason why the allocations occured.
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.