We've been seeing a kernel panic under high traffic conditions when trying to reorder/mode-change the branches on a unionfs mount where the base branch is NFS. The error message in the panic seems to indicate that it's attempting to unmount the NFS partition, but that doesn't make a lot of sense.
Information about the issue including kernel panic logs and the versions of kernels and unionfs we've tried is available at https://bugzilla.mozilla.org/show_bug.cgi?id=402754 Some basic information about what we're trying to pull off and how it's set up is at http://www.justdave.net/dave/2007/10/22/stagemo-shift-reloaded/ with the main difference from that description being that the chroot environment directories are now bind-mounted on top instead of as a branch within the unionfs mount, because of issues I'm assuming having to do with the /dev and /proc directories being buried inside a unionfs layer (I imagine copyup doesn't play well with /dev nodes :) ) I have a test case set up where I can reliably reproduce this after a few hours of copying data in and out into the staging area, so if anyone has patches to try I can find out in a day or so if they help. :) Thanks in advance for any help! -- Dave Miller http://www.justdave.net/ System Administrator, Mozilla Corporation http://www.mozilla.com/ Project Leader, Bugzilla Bug Tracking System http://www.bugzilla.org/ _______________________________________________ unionfs mailing list: http://unionfs.filesystems.org/ unionfs@mail.fsl.cs.sunysb.edu http://www.fsl.cs.sunysb.edu/mailman/listinfo/unionfs
