On Tue, 2005-06-28 at 02:49 +0200, Christian Kujau wrote: > hi, > > today my desktop machine locked up, because i did something stupid [1]
>[1] cat /dev/port > /tmp/test, then lockup...don't ask :-) Okay, I don't know much about /dev/port, so I guess this can lock up the machine, but I don't see why it would mess up the file system. I'm going to try this on a test machine to see if I can recreate this problem. > i rebooted, but jfs_fsck-1.1.6 (from ubuntu linux) found corruptions and > mounted the fs readonly (sadly the bootprocess continued and so the fsck's > output of the first corruptions is gone). fsck would not have mounted the file system read-only. More likely, fsck fixed some problems, and the kernel found something it didn't like and remounted it read-only. > i went to runlevel 1, fsck'ed > again but now the fs seemed clean. i tried to remount -o rw, but got this > error: > > ERROR: (device sda5): XT_GETPAGE: xtree page corrupt > > ok, remounting rw did not succeed, running jfs_fsck again revealed some > corruptions: > > File system object FF4098 is linked as: /etc/mailcap > File claims cross linked block(s). > cannot repair FF4098. Will release. > File system object FF5980 is linked as: /etc/mtab > File claims cross linked block(s). > cannot repair FF5980. Will release. > File system object FF37134 is linked as: /tmp/.gdmVQXJud > The path(s) refer to an unallocated file. Will remove. > File system object FF290840 is linked as: /tmp/.gdm_socket > The path(s) refer to an unallocated file. Will remove. > File system object FF290845 is linked as: /tmp/.X0-lock > The path(s) refer to an unallocated file. Will remove. > > running fsck_jfs once more (with -f) made me believe everything was ok. > i fetched the latest jfsutils (1.1.8), rebooted, the boot-process again > says something about "permission denied" when trying to cleanup /tmp > during boot. i checked again with fsck_jfs, found again a few corruptions, > a 2nd run stated the fs is clean now. > > how comes there are corruptions when fsck says the fs is clean? and: are > the above messages signs of trouble ahead? I haven't heard of this happening before. Normally, once fsck fixes the file system, it's okay unless there is something else going on, like i/o errors. I think the problem might be that since fsck is running against a currently-mounted partition, that it is finding problems with files that are currently opened. It deletes those files on disk, but the kernel still has the files open. If this is the case, I'm not sure how to address it. If the file system is mounted read-only, fsck has no way to know what files are currently open by the kernel. If it could, it could refuse to allow a read-write remount until the system is rebooted. > right now i'm running X in the > same machine, so problems right now. but the corruptions did cost me some > files in /etc too (mailcap, network/interfaces, maybe more...). I'm guessing you're back to a consistent state now. You could make sure by fsck'ing the file system from a rescue cd. Also, if you see the problem again, you could reboot after running fsck without remounting the partition read-write. > more info right here: http://nerdbynature.de/bits/prinz64/2.6.12-mm2/jfs/ > > thanks, > Christian. -- David Kleikamp IBM Linux Technology Center ------------------------------------------------------- SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click _______________________________________________ Jfs-discussion mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/jfs-discussion
