Date: Mon, 19 Dec 2005 07:17:12 -0600 From: Eric Anderson <anderson@centtech.com> To: Lukas Ertl <le@freebsd.org> Cc: freebsd-fs@freebsd.org Subject: Re: filesystem full - freebsd 5.3 Message-ID: <43A6B2D8.8000907@centtech.com> In-Reply-To: <20051219132739.R28071@pcle2.cc.univie.ac.at> References: <200512191220.jBJCKDI8037706@lurza.secnetix.de> <20051219132739.R28071@pcle2.cc.univie.ac.at>
next in thread | previous in thread | raw e-mail | index | archive | help
Lukas Ertl wrote: > On Mon, 19 Dec 2005, Oliver Fromme wrote: > >> > > Perhaps more likely, he was trying to allocate full-size blocks, and >> > > the only things available were fragments. The output from df >> doesn't >> > > distinguish between the two types of available space. You can use >> > > dumpfs(8) to do that. >> > >> > This version seems more likely for me. >> >> In the situation give, I think it is rather unlikely. > > > We have to cope with the same problem here. It's a 662GB filesystem > used for Cyrus imapd mail folders. 55GB free space, plenty of free > inodes, and yet we get "filesystem full" messages. If we remove some > mail folders (postmaster double bounce stuff, thousands of mails per > dir), the kernel stops complaining about a full filesystem (until it > runs out of $factor_x again). > > We are now moving to a new machine, where we will split up the large > filesystem to smaller ones. This is interesting - can you touch a new file? Also, can you post (http please) your dumpfs output also? I realize it might be large. The most interesting pieces to me are the superblock and cylindar summary info, plus maybe the first 100 cylindar group info. Eric -- ------------------------------------------------------------------------ Eric Anderson Sr. Systems Administrator Centaur Technology Anything that works is better than anything that doesn't. ------------------------------------------------------------------------
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?43A6B2D8.8000907>