Date: Sun, 11 Oct 2009 16:38:17 -0700 From: David Ehrmann <ehrmann@gmail.com> To: Gavin Atkinson <gavin@FreeBSD.org> Cc: daichi@FreeBSD.org, freebsd-current@FreeBSD.org Subject: Re: Strange md/unionfs issue Message-ID: <4AD26C69.4040202@gmail.com> In-Reply-To: <20091011133411.A80896@ury.york.ac.uk> References: <4AD0E286.9030508@gmail.com> <20091011133411.A80896@ury.york.ac.uk>
next in thread | previous in thread | raw e-mail | index | archive | help
Gavin Atkinson wrote: > > On Sat, 10 Oct 2009, David Ehrmann wrote: > >> I have a system running on a flash disk. To prevent it from wearing >> out quickly, I don't run with a swap and mount /tmp and /var from >> memory. Because /var needs some directories and contains files I >> probably should save from time to time, I mount it from flash, mount >> /var to /var_real so I can access the underlying files, and then I >> mount an md device on top of /var as a unionfs. Initially, it seems >> to work, but seconds later, the unionfs seems to fail. > > When the mount "fails", is it still shown in the output of mount(1)? The output of mount is the same; it reports that a md-backed unionfs is on top of /var. > > It appears that this problem is the same as is reported in PR > kern/131441, which has so far not had any activity. It appears that > this bug has existed since at least 7.1-RELEASE. It does seem to be the same bug (or similar). With mine, I mounted a unionfs on top of a mount point that a nullfs was based on, not the nullfs's mount point. I think the bug in kern/131441 was with a unionfs over a nullfs. This probably means md's off the hook as the culprit, and the same thing would happen with ufs on /dev/foo.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4AD26C69.4040202>