Added value partitioning on VPS

OldskoolOrion

OldskoolOrion

Hi,

I'm wondering what the added value is of making more partitions on a VPS besides the root and a well-sized swap partition and perhaps a third partition to (temporarily) write file-based/img based backups to in addition to the snapshots of the virtual machine itself.

Assuming the install itself is a simple server, only used as web / mail / ftp-server, with only virtual accounts added for mail and ftp users. The only real user account being able to ssh is the management-account.
What would be the added value in creating the partition-based layout if there's scripts in place run by cron to clean up not needed stuff in portstree, tmp, etc.?

It's all hosted on the SSD based storage, so there's no difference in IOPS for parts of the install.

Any thoughts ?
 
Stuck mail could fill up the filesystem. FTP users could upload large files, filling up the filesystem. A run-away web process could fill up the filesystem. There are several reasons why splitting up the filesystems would be a good idea. Performance usually isn't one of them.
 
Thank you! This confirms my view that it's still better to do, than not to do, even if it's just a fail-safe for "just in case of" scenario's (to be honest stuck mail didn't even enter the discussion yet, but certainly helps my pov :p).

Even tho monitoring is in place, I've encountered enough situations, something went horribly wrong at a time where admins where not able to intervene instantly.
 
OldskoolOrion said:
Even tho monitoring is in place, I've encountered enough situations, something went horribly wrong at a time where admins where not able to intervene instantly.
Click to expand...
Yep. I just had to fix something like that on one of our core servers. An admin had a script running that saved the output of arp -a to a file. Every 5 seconds. This script had been running since October and used up a whopping 120 GB of a 130 GB filesystem.
 
hahah :D
in a disturbed way it's funny.. i'd say 9 out of 10 times, it's not a horribly wrong configured service or whatever.. 9 out of 10 times it's an admin screwing himself or a colleague by just forgetting a tiny detail.. like stopping the logging after being done investigating something ;-)
 
You must log in or register to reply here.
Back
Top