Today morning my home machine does not boot properly (was working yesterday evening). I suspected XFS on rootfs and it was that. Booted into /bin/sh and started xfs_check -f /dev/hda1 followed by xfs_repair -f -d /dev/hda1. There were some problems but looks like it solved them so I rebooted. Again system fscked.. Another boot into /bin/sh and another xfs_check/xfs_repair combo and again nothing.
Booted into single user and then SSH connection to other machine to get help on #xfs channel on freenode. ‘dtm’ suggested me to run badblocks on disc to check does hdd is ok. So I installed Debian ‘sarge’ on my swap partition and started badblocks -v -n /dev/hda1 — 0 bad blocks found.
So I decided to reformat partition and reinstall whole system. Mounted old rootfs into /mnt/, chrooted into this to get list of installed packages with dpkg -l and archived whole partition with tar. Then umount, mkfs.xfs -f /dev/hda1 and restoring from tarballs.
Next step: apt-get install --reinstall all-installed-packages is in progress now:
Need to get 389MB/676MB of archives. After unpacking 0B of additional disk space will be used.
I hope that after this I will have working system again. System which will just works not like I had in last month when from time to time some files were missing from rootfs (kernel module or awk or other thing) but usually problem disappeared after reboot…
Maybe it is time to change something in machine — but how to find out what…