File System on Linux: Set up for large Elasticsearch

Question

I am setting up a single 750GB elasticsearch node (for logstash) on CentOS 6.8

Should I just set it up as a single large disk (boot and everything in one partition)?

Or should I do the :

/boot   # small, 15GB or so
/ everything else  #750GB

Or ?

You should do whatever best works with your deployment system, your backups, and however you have your disks set up. Beyond that, it's a matter of preference or specific requirements of a product, of which this one has little. — Spooler, Mar 25 '17 at 00:50

score 3 · Accepted Answer · answered Mar 25 '17 at 00:51

3

No matter the situation, it is nearly always recommended to segregate your application's storage from the rest of your server.

I typically create three filesystems:

/
/var/log
...and a third application-specific filesystem, mounted wherever it makes sense.

This way you minimize collateral damage if any of your filesystems fill up.

To answer your question directly, yes, I would recommend segregating your ES store from the rest of your server.

answered Mar 25 '17 at 00:51

EEAA

Thank you! The next question is (sorry for the noob aspect), what is the process to accomplish this? I know how to install linux, and to add the app-specific file system, but I find that a stock install of linux already has /var/log , so how do I replace that with a diff volume? – Jonesome Reinstate Monica Mar 26 '17 at 05:08

1 Answers1