We were working on the cluster of 2 data nodes(8 GB of each) + 1 master node(8GB)(I/O Optimized template), with data size of 7.7GB (16GB with replication).
We have only 1 index with 5 shards. Cluster used to always on 75-85% JVM pressure.
Later, we decided to upgrade the cluster and changed the data node RAM to 16 GB for each. Data size, indexes and shards still remains the same.
Still, the cluster shows JVM heap pressure of 60 to 70%. Is it normal or expected behaviour?
The JVM starts GC at 75% so 60% to 70% is generally not a problem as long as it drops when it reaches 75%.
Yes, that's correct behaviour.
But my questions is even after scaling to double the size. JVM pressure should be in the range of 40-55%, instead of 60 to 70%.
Does this has any thing to do with the write or indexing operations frequency? Because we have write heavy application, where the single document gets updated multiple times.
It will always go up to 75% before reclaiming. It is the level it goes down to after this that is important, not the current value.
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.