Unassigned primary shards. reason:NODE LEFT



Our ES cluster health is often red due to unassigned primary shards.
The unassign.reason for this is "NODE_LEFT" (See details below)
Does this mean that the node left the cluster while it was being assigned the shard?
When the node comes back, how come the unassigned shards are not reassigned back to that node automatically?

curl localhost:9201/_cat/shards?h=index,shard,prirep,state,unassigned.reason,usassigned.for,unassigned.at | grep UNASSIGNED

index1 3 p UNASSIGNED NODE_LEFT 2016-12-12T00:03:36.663Z
index1 3 r UNASSIGNED INDEX_CREATED 2016-12-12T00:01:03.944Z
index1 0 p UNASSIGNED NODE_LEFT 2016-12-12T00:04:59.915Z
index1 0 r UNASSIGNED INDEX_CREATED 2016-12-12T00:01:03.944Z


(Mark Walkom) #2

What version are you on?
Have you checked the logs on the nodes that have left?


I am using version 2.3.3. Yes, I've checked the logs of the node (I have debug turned on), but there is not much information besides the node dropped out of the cluster due to ping timeout, did not see anything regarding the shard.

(Mark Walkom) #4

Perhaps if you can post the log with the disconnect and things before and after, we can highlight something?

(system) #5

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