3

I run a CentOS vps to host a few websites with cPanel/WHM but every now and then, my websites stop working because it can't establish a connection to the mysql database.

Looking at the messages log, it shows the following:

Oct 16 04:03:07 s97-74-228-8 kernel: [54180237.755824] Out of memory in UB 88299: OOM killed process 413 (spamd child) score 0 vm:232540kB, rss:100696kB, swap:0kB Oct 16 04:03:08 s97-74-228-8 kernel: [54180237.863552] Out of memory in UB 88299: OOM killed process 1060 (spamd child) score 31 vm:231516kB, rss:99536kB, swap:0kB Oct 16 04:03:08 s97-74-228-8 kernel: [54180237.888621] Out of memory in UB 88299: OOM killed process 1061 (spamd) score 31 vm:231516kB, rss:99160kB, swap:0kB Oct 16 04:03:08 s97-74-228-8 kernel: [54180237.898954] Out of memory in UB 88299: OOM killed process 25084 (spamd) score 0 vm:231516kB, rss:100204kB, swap:0kB Oct 16 04:03:08 s97-74-228-8 kernel: [54180238.007397] Out of memory in UB 88299: OOM killed process 29642 (mysqld) score 0 vm:7849012kB, rss:62860kB, swap:0kB Oct 16 04:03:08 s97-74-228-8 kernel: [54180238.165135] Out of memory in UB 88299: OOM killed process 29450 (named) score 0 vm:1786256kB, rss:31192kB, swap:0kB Oct 16 04:03:11 s97-74-228-8 kernel: [54180241.036672] Out of memory in UB 88299: OOM killed process 1191 (mysqld) score 0 vm:1662420kB, rss:42300kB, swap:0kB Oct 16 04:03:12 s97-74-228-8 kernel: [54180242.761161] Out of memory in UB 88299: OOM killed process 1362 (mysqld) score 0 vm:807072kB, rss:40124kB, swap:0kB Oct 16 04:03:16 s97-74-228-8 kernel: [54180246.167647] Out of memory in UB 88299: OOM killed process 1434 (mysqld) score 0 vm:2188788kB, rss:43808kB, swap:0kB Oct 16 04:03:31 s97-74-228-8 kernel: [54180261.056050] Out of memory in UB 88299: OOM killed process 1562 (mysqld) score 0 vm:7980592kB, rss:59692kB, swap:0kB

After my site goes down, I ssh in but I only see at most about 446MB/3072MB used.

Not really sure what to try to be honest, any help would be greatly appreciated :)

Alex

Improve this question
3
  • 4GB in an email+anti-spam+vhost for customers+BIND+SQL server in the same box evidently is not enough. I bet you also have webmail there...It would be easier besides giving RAM temporarily to it, separate the services in more than one machine. Commented Oct 16, 2016 at 13:30
  • This is the same argument that I see people using, but what I find truly incredible is that when you have Windows machines with the same 4GB of memory, they simply work, there are never these kinds of problems on Windows side of things. Linux is terribly behind when it comes to memory management and the OOM Killer in particular is specially awful. Disk swapping code was never updated since the 90's either, it's completely obsolete and can cause premature death of disks due to the constant unnecessary I/O it performs. Commented May 19, 2021 at 10:03
  • I've just bought a new machine and updated to Debian, I'm having OOM and swapping slowness, for a brand new machine in 2021. The weird thing is that I don't experience the same OOM errors and aggressive swapping when I'm using Manjaro, on the same machine. I'm currently researching what Manjaro does different about this particular area, so I can replicate to other distros. Commented May 19, 2021 at 10:05

1 Answer 1

Reset to default
0

You can create a custom my.cnf at https://tools.percona.com. I've had success in preventing OOM's by using their configuration wizard for MySQL. Be sure to allow enough RAM for the OS, CPanel and whatever else you have running on the system.

Improve this answer

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.