Kernel 2.67.17.1 is hanging I/O under AMD64 [Was Re: Kernel 2.6.17-rc5 under amd64 may be hanging I/O]
The problem,at least I think it's the same problem, also occurs in kernel 2.6.17.1.
The relevant output has been posted online at
http://www.kennedy.aecom.yu.edu/misc/drbdstall2.htm
Quoted 37 lines
/ 2006-06-08 20:34:10 -0400
\ Maurice Volaski:Quoted 14 lines
It appears that kernel 2.6.17-rc5 under amd64 may be hanging I/O processes spontaneously at random.
Our setup here uses drbd (ver. 0.7.19), which is a network RAID kernel module, and initially I was fscking (ext3) filesystems, which are on drbd devices, and the fsck just stopped spontaneously on the two of
them.Today, I tried copying a directory on the command line with cp -a on this computer (on a drbd-managed device) and then in mid-copy I tried to abort the process with control-C. It did not abort. I tried killing it with kill and then with kill -9. It turns out that the process had died, but is still left in ps.
Quoted 9 lines
Just by happenstance,
Anyway, I tried bringing down the peer and bringing it back up and it stalled.
I also had an emerge sync (i.e., the Gentoo update mechanism) going
and it too got stuck, but it doesn't, or at least shouldn't affect drbd disks, implying this is not a drbd bug, but a kernel bug.well. it says below that emerge hangs is
drbd_al_begin_io, so it is at least drbd related, too.Quoted 4 lines
Quoted 1 line
* what are the numbers in
/proc/drbdThis is how it appears long after the copy hang and I stopped and just restarted drbd on the peer. The output from then hanging computer:
thanks, that might help in debugging things. I'll have a look, maybe I can find something in there.
Signature · 7 lines
Maurice Volaski, [email protected] Computing Support, Rose F. Kennedy Center Albert Einstein College of Medicine of Yeshiva University _______________________________________________ drbd-user mailing list [email protected] http://lists.linbit.com/mailman/listinfo/drbd-user