drake127 wrote:
I managed to crash 2.6.28.3-x86_64-linode5 during recursive chown Gentoo portage on loopback device formatted to reiserfs twice in row. Since it is production machine, I'd like not to test it again without better ways to describe the issue.
I did something like this:
Code:
mv /usr/portage /usr/portage.tmp
losetup /dev/loop0 /usr/portage.img
mkreiserfs /dev/loop0
mount /dev/loop0 /usr/portage
cp -a /usr/portage.tmp/* /usr/portage
# Everything worked fine so far
chown -R portage:portage /usr/portage
And whole linode crashed. There was kernel panic report on console screen but unfortunatelly was not complete and I did not discovered it in logs.
Can somebody try similar procedure or tell me what I am doing wrong or tell me how to provide more information?
I tried this on my linode with the same kernel and could not recreate your problems. I didn't do it directly on /usr, but made a separate lv for testing. I tried reiser on reiser, reiser on ext3, and ext3 on ext3, and in all cases I could copy /usr/portage/* and chown -R the mounted loopback without errors.
Code:
lvcreate -L1G -ntest vg
mkreiserfs /dev/vg/test
mount /dev/vg/test /mnt/tmp
dd if=/dev/zero of=/mnt/tmp/portage.img bs=1K count=500K
losetup /dev/loop0 /mnt/tmp/portage.img
mkreiserfs /dev/loop0
mkdir /mnt/tmp/portage
mount /dev/loop0 /mnt/tmp/portage
cp -a /usr/portage/* /mnt/tmp/portage
chown -R portage:portage /mnt/tmp/portage
I'm not sure what is different, but it seems stable for me.
FAQ
Search
Members
Register
Login [ Anonymous ]
