Thomas Kuther wrote:
On Sun, 13 Nov 2005 13:29:25 +0100
Artur Makówka <[EMAIL PROTECTED]> wrote:
it crashed today morning, so 2.6.13mm is not much better than 2.6.14.
My fs is heavly used (but not overloaded) by many apache process or
ftp process or postfix process.
I have free hosting server and i think i will have to close this
thing becouse for over a month its often downtime everyday. Could you
please tell us if there is any idea of how to fix it? i would move to
another more stable fs but its too much data..
hmm, wondering which I/O scheduler you guys are using.
I had the same slowdowns and soft lockups till 10 minutes ago, when i
switched from cfq to anticipatory I/O. I'm currently usig 2.6.15-rc1 +
reiser4 from 2.6.14-mm2
I was in the middle of a big gentoo'ish emerge, desktop almost
unusable. but now with anticipatory all seems fine so far.
the compile seems a lot faster. no soft lockup trying to use vim or
locate... hmmm
regards
tom
i was using anticipatory I/O as default in debian. i just did echo cfq >
/sys/block/hda/queue/scheduler. Hope its enough to change it.
no noticed effects yet, i will give report, if anything changes.
