SOLVED Partimage/libuclibc crash

  • EDIT: Ok, nevermind, I should create the task through the webinterface instead of writing the image manually with partimage. That way it works. I’ve no idea why I crashed in debug mode when I ran partimage, but that isn’t important now anymore 🙂
    Feel free to delete / mark as solved.

    ** original message below **


    When I boot fog in debug mode because I want to create a new image[*], and then run partimage it crashes at 0% after hanging for 5-10 seconds:
    partimage[xx]: segfault at 0 ip b75ee98a sp bfa67c84 error 6 in[…]
    Segmentation fault

    I hope this is a simple issue / oversight on my end. That said, I searched the forum and wiki but didn’t find any others with this problem.

    I already tried using another 3.3 kernel (from another forum thread), and compiled my own custom kernel, so tried 3 kernels in total. But this does not resolve the issue. Seems to me that this is unrelated to the kernel.
    Also ran memcheck on the box. Nothing.
    Also ran all of the above on completely different hardware (first one was 1 month old, second one 4 years), same result.

    All other (though, limited) commands work. Including dd if=/dev/zero of=something bs=1M count=1000 over the network NFS mount.

    The crash of above happens both when trying to save to a NFS share and just local to /tmp.

    This is version 0.32 of fog.

    If you are unsure what it is, feel free to give any suggestions as to how I can trace this issue or what else I can try.

    [*] there’s no GUI/interface for image creation right? I could have overlooked this, but didn’t find anything in the wiki. It’s okay as we’ve been using a Linux shell for the past 5+ years for this process, I was just slightly surprised.

  • The GUI is the most appropriate place to schedule/setup the tasking.

    That said, it should still be possible to setup tasking directly via the command line within the client you’re trying to perform the task.

    However, it’s not very simple as it is unlikely your system has enough RAM to store the image. Process should be:
    Mount /images/dev
    Make a directory within the mounted /images/dev to store your image
    make a backup of your mbr and store it in the location.
    make file input file output interface (mkfifo)
    Translate and obtain image through that fifo file to the storage location (Partclone or Partimage)
    You’ll likely need to do this for all partitions you need the data from.
    Once all of that is complete you’re fairly safe to reboot the client machine.
    You’ll then need to copy or move the /images/dev/<FOLDERNAME> to /images/<IMAGENAMEPATH>

    THere is a normal file format standard for fog. (Assuming multi-part image type)
    d{disknumber}.mbr (e.g. d1.mbr d2.mbr) Is the mbr naming convention for the disk.
    d{disknumber}p{partitionnumber}.img (e.g d1p1.img d1p2.img d2p1.img d2p2.img) Is the Naming of the actual image data.

    For resizable images the name is:
    (For standard base install of windows 7)
    rec.img.000 (the 100MB recovery partition)
    sys.img.000 (the main data patition)

    If there is no 100MB recover partition and you have only the one partition the only file needed is

    Hopefully this helps, but it’s not very straight forward. It’s much easier to use the GUI to schedule the deployment job you need for both Upload and Download tasks.