When I was first trying deploying elasticaserach stack using ECK on Google Cloud Kubernetes in Auto Pilot mode. I could not perform only one thing from the documentation which was setting
command: ['sh', '-c', 'sysctl -w vm.max_map_count=262144']
on the elasticserach nodes
Do I really need this command to run on elasticserach pods and without that, I can expect some unwanted unexpected behavior of the cluster, or will there be just a max cap of RAM that the node can properly use? If so what is that value?
My question was raised from my last experience trying to deploy elasticserach cluster using ECK and Google Cloud Kubernetes in Autopilot mode. When I tried to deploy this configuration you can notice that I kept privileged: false, because It's not allowed to run privileged pods as this says(This generated some error about command but was running just without it).
I still want to use autopilot mode for bigger abstraction, less overhead, and cheaper solution.
Summering, my question is what are the limits that I should expect(like index size or maybe max_heap setting) leaving vm.max_map_count as the default value 65536?
I have read the documentation one more time and I came to a conclusion that these privilege pods do not matter if you choose for example niofs storage type.
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.