I'm not quite sure whom to get ahold of here, but I just had to reboot
my server because it went off into never never land and never came
back. What did I do? I created a level1 raid out of two loopbacked
files on the same filesystem for testing purposes.
# $CWD is on a freshly-formatted JFS partition
for i in 1 2; do dd if=/dev/zero of=raid.${i} bs=1k seek=$( echo
1024*1024*20 | bc ) count=0 conv=sync; done
losetup /dev/loop4 $PWD/raid.1
losetup /dev/loop5 $PWD/raid.2
mdadm --create /dev/md99 --level=1 --bitmap=internal --raid-devices=2
/dev/loop4 --write-mostly /dev/loop5
Shortly after the above command, I saw /proc/mdstat rebuild speed drop
quite rapidly below 10MB/s and then the entire machine became
completely unresponsive. Oh, the disk was making plenty of noise but I
gave it 5 minutes and only the console responded, by echoing back my
characters.
Otherwise, even to magic sys-rq, it appeared completely unresponsive
although not "dead".
Any ideas?
I've tried the same using the *same* disk/partition but formatted
using ext3 and did not have the same issue.
openSUSE 11.0, fully up-to-date, kernel 2.6.25.18-0.2-default on x86_64.
I've generally had pretty good luck with ext3 and jfs over the years,
but increasingly I'm worried that jfs is no longer under development.
:-(
--
Jon
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
