.security shard unassigned

Just starting to implement security in our proof of concept ES 5.1.1 cluster. .security index is yellow with 2 unassigned shards. 25 data-ingest only nodes, 3 master only, 1 client only, (1 kibana only). The security index has 1 primary and 24 replica but 2 are unassigned. How do I correct this?

Why do you have so many replicas?

I dont know why i have so many replicas. I only noticed when we started setting up users. What would be 'normal' replicas for security? How do i get to that? Cluster is working as expected, including security. Just trying to see it green again.

You probably only need one replica, that's default, but it depends on your failure/recovery requirements.

You may also have scaling replicas enabled?

As you suggested GET .security showed "auto_expand_replicas" : "0-all" (FYI, the default for .security used by the Security module. Are you sure that 0-all isn't required by the Security module??

Assuming the best (yikes), The following changed that to false, reduced the replicas to 3, and the cluster is green.

PUT .security/settings
{ "index" : { "number_of_replicas" : 3, "auto_expand_replicas" : "false" } }

Important, I never set auto_expand_replicas to 0-all, I think that was the default for .security module !!

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.