Elasticsearch tuning

[rferebee]

Ok, fair enough.

Regarding CentOS 6 vs 7. I am very close to replacing LSCC2 with a CentOS 7 machine. It is in the process of being configured.

I think we can probably close this out since it's such a rare occurrence. I will say though that whenever Elasticsearch does get hung, I am unable to restart it without getting the memory error I provided. That happens every time.

Perhaps with a new VM running CentOS 7 and the same size swap it won't happen anymore.

Thank you!

[mbellerue]

Yes, I think the reason it's giving you that error is because it's failing to stop the hung instance of Elasticsearch. So with that hung instance taking up 50% of your memory, there's no way that you could spin up another instance unless there was nothing else running at all.

You could probably find the process by doing

Code: Select all

ps -ef | grep -i elasticsearch

Then find the java process running it, and kill the PID. If you're comfortable doing that. It may be faster than a straight reboot of the VM.

[rferebee]

Ok, I'll try that the next time it happens.