It seems like i lost some of the text in the report above, the full sentence 
should be;
While copying data from old /home-partition on one drive to new /home-partition 
on a new drive (both dm_crypt + ext4), watching video from old /home and 
Tectonicus minecraft-map-renderer starting an hourly render, disk activity 
suddenly stopped.

I have done some additional testing, using linux-image-2.6.35-30-generic
from maverick is stable.

I set up a virtualbox machine, attached a physical partition and did the
same setup as above; natty + dm_crypt + ext4. Then i configured
netconsole so i would see any errors even if the machine froze, and
started two scripts running over and over again. One compressing and
extracting the same 400MB folder of 1MB .rar-files, another copying a
70GB folder and removing the copy.

I ran this for 1½hour without errors, and then i stumbled over;
http://osdir.com/ml/linux-ext4/2010-12/msg00054.html
which made me realize that to mimic the hostmachine more correctly i should 
give the virtual machine access to all 4 cores instead of the virtualbox 
default of just one.

I left it running over the night and today all I/O on the machine was frozen, 
but no message in netconsole.
I'm gonna try this setup again to see if i can get some kind of error, with 
another fs to see if it's ext4-specific, and I'm thinking about compiling the 
kernel with the parallel dm-crypt-patch reversed, to see if that's the issue, 
even though it's reported solid;
http://osdir.com/ml/linux-ext4/2010-12/msg00020.html

Any more suggestions?

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/797127

Title:
  kernel BUG at ext4/inode.c:2082 during heavy I/O

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/797127/+subscriptions

-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to