On Wednesday 09 June 2010 09:32:40 Neil Bothwick wrote: > On Wed, 9 Jun 2010 06:03:51 +0000, Mick wrote: > > Thanks I used reiser4progs to check and repair the fs. The weird > > thing is that I had to repeat this on the /var partition, after I > > zero'ed it, reformatted it and reinstalled gentoo on it. O_O > > > > How is it possible that the same directory/file gets corrupted again > > after a reinstall? > > The first thing I'd do it set PORTAGE_TMPDIR to somewhere less important > than /var. You don't want to risk corrupting system files while sorting > this out. It would also show whether the problem was with an ebuild or > the filesystem. Use a tmpfs filesystem if you have the memory.
I had to reinstall because I didn't catch this early enough the first time and it corrupted my other system partitions. This corruption happened when I installed gcc-4.4.3-r2. After I reinstalled and while still running the LiveCD I updated gcc to the same gcc-4.4.3-r2 version. As I said, again I ended up with a corrupted '/var/tmp/portage/sys-devel/gcc-4.4.3- r2/work/gcc-4.4.3/libjava/classpath/resource/gnu/java/locale' directory. Running fsck.reiser4 --check and then --fix corrected the corruption and that was that. Before I reinstalled, the corruption must have been more pervasive because fsck.reiser4 could not fix it - a simple reboot would bring it back. So, my conclusion from this sad story is that on my system reiser4 was attacked by gcc! O_O I'd be interested to know if any other gentoo systems running a separate /var partition on reiser4 noticed this, in which case I'd file a bug for it. If I'm alone in having this problem then I'll wait until it happens again before I open a bug report. -- Regards, Mick
signature.asc
Description: This is a digitally signed message part.
