Hi, guys!
I have some problem with memory consumption, I jope you will help me with it.
OS: RHEL 7.3
Elasticsearch 5.4
RAM: 8GB (real 7.6GB)
Clear install of EL
Nothing in index
Nothing else run on machine
Nothing request EL
JVM Heap set to 4GB
swap off
all descriptors, threads tuning by EL guide
And I meet something strange as for me
* ps -eo pid,ppid,cmd,comm=,rss,vsz,%mem,%cpu --sort=-%mem | head
PID PPID CMD RSS VSZ %MEM %CPU
46433 1 /bin/java -Xms4g -Xmx4g -XX java 4661496 6906888 58.2 0.2
VSZ show that virt memory consumption is ~6.5GB
* free -m
total used free shared buff/cache available
Mem: 7815 4575 837 96 2401 2697
Show that only ~4.5GB is used
* curl -X GET '127.0.0.1:9200/_cat/nodes?v&h=id,ip,port,v,m,ram.current,ram.percent,ram.max'
id ip port v m ram.current ram.percent ram.max
aRwE 10.48.5.59 9300 5.4.0 * 6.8gb 89 7.6gb
EL think it use about ~6.8GB (89%MEM)
* top
KiB Mem : 65.6/8002828
KiB Swap: 0.0/2097148
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
46433 elastic+ 20 0 6906888 4.447g 125296 S 0.0 58.3 0:40.56 java
Top said that ~6.8GB (58.3%MEM) but from simple math 6.8GB/7.6GB * 100% = 89% (like in EL request)
Where the truth? EL virt memory include some buff/cash from system? If it isn't true, why that much memory consumption (~4gb HEAP) + 2.5 GB [for what?], if no index and data...