Date: Mon, 21 May 2007 16:39:39 -0400 From: Kris Kennaway <kris@obsecurity.org> To: "Rick C. Petty" <rick-freebsd@kiwi-computer.com> Cc: freebsd-fs@freebsd.org, Kris Kennaway <kris@obsecurity.org> Subject: Re: VERY frustrated with FreeBSD/UFS stability - please help or comment... Message-ID: <20070521203938.GA81333@xor.obsecurity.org> In-Reply-To: <20070521203556.GA84210@keira.kiwi-computer.com> References: <475187.33232.qm@web63006.mail.re1.yahoo.com> <4651F897.9010201@infidyne.com> <20070521195634.GA80608@xor.obsecurity.org> <20070521203556.GA84210@keira.kiwi-computer.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, May 21, 2007 at 03:35:56PM -0500, Rick C. Petty wrote: > On Mon, May 21, 2007 at 03:56:34PM -0400, Kris Kennaway wrote: > > > > > > The one thing I *have* experienced is filesystems going nuts after > > > filling them. Specifically I end up with a negative amount of space > > > free, and as soon as you release more space df -h quickly reports the > > > amount of free space dropping (you can see it interactively churning for > > > a few secs) until it settles at some negative amount again and you're > > > back to square one. This has happened once or twice in the past, but not > > > recently. Not sure which release it did happen on; might have been 5.x. > > > > Yeah, this was an old bug in the 5.x timeframe. As far as I recall it > > was fixed at some point in the past. > > Hmm, this recently happened to me in 6.2-STABLE. I was unable to replicate > it so filing a PR was pointless. What I saw happen was that /usr filled up > completely (I caught it before it hit 100% full, but watched helplessly as > it switched from time to space optimization). I even did a "du -hd0 /usr" > and saw it only using about 20% of the file system. lsof and fstat weren't > terribly helpful. After shutting down to single-user mode, I saw it full > even though I deleted a lot of superfluous files, and du/df differences > were still present. I restarted and fsck cleaned up the 80% "used" space, > stating the superblock free maps were incorrect, which explains the > discrepency. Why it happened, I'm still baffled. OK, it may be that the bug is still around in some form, or it could be a different issue. One thing to check is whether you have a snapshot active, because this will cause a very similar behaviour. Kris
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20070521203938.GA81333>