Elastic search takes up a lot of virtual memory under Linux

[zekaifeng (kaifeng ze)]

Our Elasticsearch version is 7.10.2. It runs on Linux system. The JVM memory allocation is about 3G, and the node memory is about 8G. It is a three node cluster system. The cluster has eight shard, each of which is about 20G. In the no-load state, the virtual memory occupies 45g. Why?

[Christian_Dahlqvist]

Elasticsearch uses some off-heap memory in addition to the heap, but also makes use of the operating system page cache to cache frequently accessed files in memory for better performance. I suspect this is what is showing up as memory usage. This is generally not a problem as the operating system will reclaim memory from the cache if needed, but make sure you have got swap disabled.

[zekaifeng (kaifeng ze)]

@Christian_Dahlqvist

Elasticsearch uses some off-heap memory 8n addition to the heap, but also mKes use of the operating system page cache to cache frequently accessed files in memory for better performance. I suspect this is what is showing up as memory usage. This is generally not a problem as the operating system will reclaim memory from the cache if needed, but make sure you have got swap disabled.

I queried the virtual memory usage details of elasticsearch process and found that most of them are occupied by shared, and swap has been disabled.