i have a cluster with 3 nodes and too many index (about 13400 with 1 shard per index, 26800 shard), the cluster fails to do resharding, because the heap memory becomes full (10g on 16 g total), i tried to delete index but i have an error of timeout (also with timeout_master option set to 5m), is there any way to force the elimination? I do not know how to restart the cluster...
Yeah, that is a lot!
You can shell into each of the machines and restart them. There is a setting,
gateway.expected_nodes that you should set so the startup waits before it freaks out about not all the nodes being up.
Frankly as this point I'd bump the heap on all the nodes to 13GB, delete the indexes you can get away with deleting, and then drop the heap again, restarting the cluster both times I changed the heap.
I expect the cluster is fairly unusable with all of the timeouts.
I resolved disabling sharding allocation, i managed to delete indexes, and then I enabled sharding
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.