Hi all,
We are planning to migrate from 0.90 version of ES to the latest version.
Also we are revisiting our production architecture.
Currently we have one master and one replica of the data. Both currently
are 16G ram servers.
My question is, is it advisable to have a better configuration master sever
ie say 32G ram on master and 16G ram on a slave node?
No. Which copy of the shard is the replica and which is the slave can change all the time. Most actions take almost the same amount of CPU/memory/time on master as slave. If you find this not true for your workload your best bet is to fiddle with the allocation parameters and raise the weight of the one for masters. This will cause elasticsearch to tend to spread masters move evenly.
Hi all,
We are planning to migrate from 0.90 version of ES to the latest version. Also we are revisiting our production architecture.
Currently we have one master and one replica of the data. Both currently are 16G ram servers.
My question is, is it advisable to have a better configuration master sever ie say 32G ram on master and 16G ram on a slave node?
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.