Just to bring anyone that was interested up-to-date.
I ran the test again after compiling with kgcc as I was told that the stack
trace was borked. The problem still existed with or without xfs_freeze so it
shows that it is an interaction problem with LVM & XFS.
After talking to the LVM guys I have decided that following through with this
is a waste of time as LVM-1.1rc2 development is being dropped in favour of
development in LVM2.
Therefore, I won't take this any further.
Thanks for those that have helped and gave suggestions.
On Wed, 29 May 2002 02:12, Adrian Head wrote:
> Steve, Eric
> This is the problem that I was discussing earlier on IRC.
> I've come across a deadlock problem with xfs_freeze when used with
> reasonable I/O activity.
(Public Key available on request.)