6.0 VFS/VM over-runs. Was: Apparent strange disk behaviour in
6.0
Julian Elischer
julian at elischer.org
Thu Jul 28 21:13:48 GMT 2005
I thought the original description was good..
file data written to cache to await output to disk is allowed to grow to
the extent that running programs are swapped out.
(or at least paged out).
here's a simple experiment that takes 1 minute to show the problem.
Best with a system with 2 disk drives (or a disk and a raid in my case)
boot to single user.
mount a large partition that has some room.. hopefully several times
the space of all RAM.
.
do:
dd if=/dev/zero of=/${mountpoint}/bigfile bs=128k count=1000000 & iostat 1
notice that dd dies with "not enough swapspace" after some number of
seconds.
(there is no swap attached). also notice that until the crash, IO speed
was max for the
disk.
then add swap and do it again.
notice that at teh point where dd previously crashed, the swap disk
springs into life and disk IO on your main disk goes into the toilet.
notice that dd gets swapped out. (!!!!)
when the disk recovers enough, dd pages back in and creates more IO but
in teh mean time there is no IO
going on to the disk.. (watch the LEDs).
Scott Long wrote:
> Julian Elischer wrote:
>
>> This is an absolute show stopper for 6.0
>>
>
> Can you distill this down to a problem description that we can track?
>
> Scott
More information about the freebsd-current
mailing list