running out of swap space

[Red Squirrel]

My online server stopped responding as it was getting more and more sluggish. I thought it was a DDoS attack but turned out I ran out of swap. How does this happen and what are things I can do to prevent it? Is it a memory leak that would cause this?

I had to break my 1 year uptime record. 🙁 I actually hit 1 year a few days ago lol.

 

[mundane]

Which process was chewing up the swap?

 

[Red Squirrel]

No idea, the data center techs had to reboot it physically. Is this something that would be caused by a specific process?

 

[n0cmonkey]

What does the server do? How much swap? How much memory? Add a swap file.

 

[Kirby]

Originally posted by: RedSquirrel
No idea, the data center techs had to reboot it physically. Is this something that would be caused by a specific process?

The swap hold pages that aren't in currently in memory, so if a process needs more memory than whats in physical memory and swap, it has no place to put it. So you need more memory or a larger swap partition (or swap file). If a process is leaking memory, I would think it would definitely fill it up.

 

[Nothinman]

This isn't something you can diagnose after the reboot, you need to see what's happing as it happens.

 

[Red Squirrel]

Is there any logs I could look at, or anything to look for, like a process taking up more and more memory or something?

This is a VM/mail/DNS/web server. The web part is hardly used. There's one VM that has windows on it which runs my game server, I'm thinking maybe the VM started using more ram. The service has not been restarted in probably about a year (till recently).

 

[Nothinman]

If the OOM killer kicked in there should be something in the kernel logs as long as it didn't get lost then the box was hard power-cycled.