[REPOST] fsck segfaults trying to repair HFS filesystem

From: Mark Bartelt (mark_at_cacr.caltech.edu)
Date: 06/24/04

  • Next message: Erika Berger: "Re: NIS-Caching at HPUX 11"
    Date: Thu, 24 Jun 2004 18:44:23 +0000 (UTC)
    
    

    [ I posted this at the beginning of the month. But sadly, I got
       exactly zero replies. I figure, at the risk of being horribly
       redundant, I'll post it one more time, in hopes that somebody
       who has a useful suggestion (or better yet, somebody from HP
       with access to a non-segfaulting fsck!) might have missed it
       the first time around. If not, I guess I'll just give up and
       build a new virgin filesystem on those disks ... ]

    Hi, folks ... We have a rather long-in-the-tooth Superdome,
    running HP-UX 11.11, and one of its disks (part of a volume
    group comprising eight drives) recently developed a bunch of
    bad blocks.

    The filesystem living on the logical volume isn't backed up.
    And of course, we have a couple users who, even though they
    were aware (or should have been aware) that there aren't any
    backups, are hoping to retrieve their data.

    So, my plan was basically this:

    (1) Do an image copy of the bad disk onto an identical (but
    good) disk. For those sectors which are unreadable, write a
    sector of all-zeros.

    (2) Replace the bad drive with the good one.

    (3) Run fsck on the filesystem (it's HFS; historical reasons
    not worth delving into ;-) to resurrect it as much as possible.

    Steps (1) and (2) went just fine. Unfortunately, when I try to
    repair the moderately-garbled filesystem (several dozen blocks,
    which were bad on the original disk, are of course wrong; plus
    there are most likely other inconsistencies resulting from the
    fact that the hardware problems meant the filesystem didn't get
    unmounted cleanly), fsck gets a segmentation fault and drops a
    core file.

    One would hope that fsck, if confronted with a filesystem that
    it couldn't put back together, would at least give up the ghost
    more gracefully than _that_.

    I'm hoping that there might be some updated fsck kicking around
    somewhere, fixing whatever bug is causing it to segfault. But
    our friendly HP guy reports that the most recent version he has
    access to appears to be identical to the one we have.

    Any useful suggestions? Do any HP insiders have access to some
    non-standard (and perhaps non-segfaulting) version of fsck that
    we could try?

    Oh, and I did try using "-f" with "-F hfs" to force a mount even
    though the filesystem is dirty. But unfortunately, with things
    in this current not-repaired-by-fsck state, only around 1% or so
    of the contents can be accessed. (With the filesystem still in
    its garbled state, one gets directory read errors on most of the
    directories when trying to descend through the hierarchy.)

    Many TIA ...

    ---------------

    Mark Bartelt
    Center for Advanced Computing Research
    California Institute of Technology
    Pasadena, California 91125

    626 395 2522
    626 584 5917 fax
    626 628 3994 e-fax

    mark@cacr.caltech.edu

    http://www.cacr.caltech.edu/~mark


  • Next message: Erika Berger: "Re: NIS-Caching at HPUX 11"

    Relevant Pages

    • Re: Damaged file system
      ... > tree structure of my second disk. ... > Windows and a reformatting of the whole disk. ... fsckis the name of the unix FileSystem ChecKer. ... I suspect that fsck created these files because the disk / filesystem is ...
      (comp.os.linux)
    • fsck segfaults trying to repair HFS filesystem
      ... group comprising eight drives) recently developed a bunch of ... The filesystem living on the logical volume isn't backed up. ... Do an image copy of the bad disk onto an identical (but ... Run fsck on the filesystem (it's HFS; ...
      (comp.sys.hp.hpux)
    • Re: fsckd - FIXED
      ... filesystem errors on boot, with the "hit control-D to continue or give root ... to fix the errors with fsck. ... filesystem and inability to reboot. ...
      (Debian-User)
    • Re: Ignorant user overfilled /usr; strange errors followed. Fixed with fsck -y, but what exactly
      ... While attempting to follow filesystem activity (and not realising the ... X/KDE, and found a whole sequence of 'filesystem is full' errors on ... Don't run fsck on a live filesystem; ... filesystem that apparently had no free blocks. ...
      (freebsd-questions)
    • Re: UNEXPECTED SOFT UPDATE INCONSISTENCY; RUN fsck MANUALLY
      ... BIOS menu item for disabling said feature. ... on a disk will be enabled, ... that fact ignored, then the filesystem is either 1) worthless, or 2) ... that's all fsck is going to do on a UFS2+SU filesystem. ...
      (freebsd-stable)