Gentoo kernel cache?

Ok..... according to netstat -ntp I only have ~100 connections opened. Yet, I will so often get "No buffer space available" when trying to open a new socket.

What is this, and how do I solve it?

`free -m' returns "58" under the mem:free collum.. but nothing is using this memory. I have 4 GB RAM, and proccesses are only using 1 GB of it. The rest is kernel-cached according to htop. If I free the kernel cache (sync; echo 3 >/proc/sys/vm/drop_caches), it quickly fills back up.

I am running Gentoo GNU/Linux without a GUI. What shall I do?

Ask on the Gentoo forums. :)

or the gentoo irc channel at freenode

Tried both, they don't really seem to know :V

No Clue. Dirty fix is to just Crontab it till it's fixed, maybe you installed a faulty update/package recently?

Nope, didn't run any updates except on sudo which had some bug. Don't think that would help it.

The command helps it temporary *I THINK*, but I am not sure. It's not a constant bug, just happens often enough to cause problems.

I am pondering installing Arch or Ubuntu or something, unless it could be because of a crappy NIC or whatnot :/

Perhaps you didn't config your kernel correctly, or a running daemon goes FUBAR.

Dunno, I never configured it, someone at some hosting company did.

It was working fine until a few weeks ago, but nothing was changed during that time is what's stumping me.

Perhaps someone tried to (and probably failed to) root your computer? Race conditional exploits generally eat up alot of buffer space (highly unlikely but you never know). Did you try to report it as a bug on their site to see if anyone else had the same problem w/ a similar setup? Though it does sound like a kernel problem, maybe they did mess up when compiling/installing the kernel :(

Didn't file a bug report.

I think it's qemu that is memory-leaking or something. I rebooted and didn't have any problems. Have not re-ran it yet either though.