Re-indexing only one node


#1

After server re-start, it looks one of 4 node has very less index data.
I am not sure but my guess is that it has corrupted indexes.

I can see one temp file i.e tmp_265d2303723b91843eb605351950f4e5_main in the index directory for that bucket on faulty node.

I dont see such temp file in healthy node.

Is there any way to re-build index only for faulty node ?
Or any other way i can recover the indexes for faulty node?
I dont want to change view code.


All nodes in cluster in pending state after m/c restart
#2

The simplest (and most robust) solution would be to:

  1. Rebalance out the node with the corrupted indexes (or failover it if you have sufficient replicas).
  2. Re-add the removed node and rebalance.

#3

Rebalance out = remove option on UI ?


#4

Yes, if you click the remove option on the UI then the node will appear as pending rebalance. You can then initiate a rebalance to remove the node.