Is there a way to determine the number of questions per shard?
I have a 6 node cluster which was evenly balanced traffic/load wise.
When I added a 7th node and rebalancing took place the 7th server's cpu
usage went to 100%.
My quick solution was to randomly move 2 non-primary shards off to other
nodes.
Of course it was eventually rebalanced so 2 shards were moved back to the
7th node.
This has fixed the problem for now.
can you elaborate, what exactly you mean with "questions per shard"?
When you add another node to your cluster, usually a rebalancing of data
happens from some of the other nodes to the new nodem which is a resource
intensive operation. But it seems, you know, so I'd like to know more about
your case and what exactly has happened in order to find out if this is a
bug or a feature?
Is there a way to determine the number of questions per shard?
I have a 6 node cluster which was evenly balanced traffic/load wise.
When I added a 7th node and rebalancing took place the 7th server's cpu
usage went to 100%.
My quick solution was to randomly move 2 non-primary shards off to other
nodes.
Of course it was eventually rebalanced so 2 shards were moved back to the
7th node.
This has fixed the problem for now.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.