Hi, I see (presumably) this problem too, though its taken a long time to even characterise. It's followed me through 2.6.10 to 2.6.12-rc1, I've not tried the 2.6.11 kernels that were released after 2.6.12-rc1 hit testing.
I don't build mysql (though this is a build machine), but the problem manifests as a slow leak that gradually fills my swap until after 3 - 5 days, even simple operations begin to thrash. I also believe it is being lost in the kernel somewhere fairly firmly now, and I've had a hard time pinning down what if anything in userspace specifically aggravates it. I can win back some fairly large chunks of swap by restarting long running memory churners, spamd and mozilla seem to be particularly profitable in that respect, but essentially it means a 4 day reboot cycle on the machine with this problem. :( Is no one else seeing this? thanks, Ron -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
