Why is my search head cluster captain logging KV Store replication errors?

[twinspop]

The log is repeating at sub-second intervals:

2017-10-27T20:44:53.389Z I REPL [ReplicationExecutor] Error in heartbeat request to shccaptain:8191; InvalidReplicaSetConfig Our replica set configuration is invalid or does not include us

The kvstore appears to be healthy otherwise.

> curl -sku admin:password https://shcmember:8089/services/server/info | grep -i kv
<s:key name="kvStoreStatus">ready</s:key>

> curl -sku admin:password https://shccaptain:8089/services/server/info | grep -i kv
<s:key name="kvStoreStatus">ready</s:key>

What's the problem? What's the fix?

EDIT: The error started after creating a collection by running this curl command:

curl -ku admin -d name=userid https://shcmember:8089/servicesNS/nobody/alpha_search/storage/collections/config

[twinspop · Accepted solution]

Splunk Support provided this solution:

on the SHC captain, splunk stop, then splunk clean kvstore --local, accept the scary warning message, and finally restart splunk. The errors have subsided and as far as I can tell, no damage was done.

Hope this helps a future admin.

[peter_krammer]

Thanks!
Has the same problem today and changing the captain did not help.
But running this on the problematic host, solved it.

[delink]

A slightly safer way to do this is just transfer the captaincy to another member of the cluster so that it rebuilds the replicaSet configuration for mongod. No stops or cleans required.

https://docs.splunk.com/Documentation/Splunk/latest/DistSearch/Transfercaptain#Change_the_captain

[kamlesh_vaghela]

HI
can you share collection configuration?

[twinspop]

The collection was configured as above. No other settings applied. These are the defaults as far as I can tell:


[userid]
profilingEnabled = false
profilingThresholdMs = 1000
replicate = false
replication_dump_maximum_file_size = 10240
replication_dump_strategy = auto
type = undefined


[kamlesh_vaghela]

Hi

Can you please try below stanza?

[userid] 
replicate = true
enforceTypes = false

I have removed other configuration bcoz those have assigned to default values. Make sure about configuration would same in entire cluster.

Thanks

[twinspop]

I should say that i have many other collections running, with exactly the same creation procedure. None triggered this error repeating. That said, I added the configs you suggested above to this single collection via the REST API. Confirmed via btool. The errors continue.