This isn't really a cPanel issue, but has helped the performance of one of our servers, so I thought I'd share. Forgive me if I posted this in the wrong place, I'm still a relative n00b in the forums.
We had a server use quite a large amount of swap. After rebooting, the server was again using a large amount within several days.
600 MB of swap! We have 4 GB of RAM in this box!
root@XXXX [/etc/init.d]# vmstat
procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
r b swpd free buff cache si so bi bo in cs us sy id wa
1 0 601852 30952 76500 928884 0 0 6 3 3 8 20 7 61 12
After combing ps aux and running top with M option (must be used interactively, top -M and top M didn't seem to work) I noticed that minilogd is using crazy RAM (61%). What is minilogd?
I went through cPanel forums, chirpy says google it. Googled. I couldn't find a definitive explanation, but from the pieces I've found:
minilogd is a logging application that starts prior to syslog on bootups. minilogd sits in queue collecting kernel messages until syslog can get up and running and take over. While it's queueing, it eats up memory and dramatically kills performance.
Upon checking, syslog was not running.
service syslog restart
As soon as I restarted syslogd, minilogd started handing off all of the messages its been collecting since the last reboot (roughly 12 days worth!).
Shutting down kernel logger: [ FAILED ]
Shutting down system logger: [ FAILED ]
Starting system logger: [ OK ]
Starting kernel logger: [ OK ]
Once minilogd was done passing off the message, the memory it was using was freed up:
WOW! 3 GB of RAM usage by a single process!
root@XXXX [/etc/init.d]# free -m
total used free shared buffers cached
Mem: 4052 1208 2844 0 67 822
-/+ buffers/cache: 318 3734
Swap: 4094 0 4093
Our average load dropped dramatically once minilogd was no longer running.
Hope this helps someone.