Announcing vmfs-tools version 0.2.0

This release should have occurred much earlier, but I wanted to work on a test suite before. Seeing how that was not going to happen any soon, I decided to defer the test suite to 0.3.0 and released 0.2.0 today.

It adds a basic support for MBR-style partition tables as they are usually found on VMFS containing LUNs (i.e. a single 0xfb typed partition), a basic fsck tool which only does a few checks and no corrections, and a better debugvmfs shell (but still with the same commands, which ought to change in 0.3.0). It also supports loading a multiple extents VMFS with missing ones (except the first one, which contains everything).

This new release has also been tested to build and work on FreeBSD 7.2, Opensolaris 2009.06 and Cygwin 1.5.25.

2009-11-23 18:49:05+0900


You can leave a response, or trackback from your own site.

11 Responses to “Announcing vmfs-tools version 0.2.0”

  1. jd Says:

    Hey, been reviewing your source code and didn’t realize their was an alternative to the Open VMFS drivers by fluidops. Great job on the coding – saw your TODO list, keep up the good work.

  2. ronnie Says:

    Any idea when the git for vmfs-tools will be back online?;a=summary

  3. glandium Says:

    ronnie: Fixed

  4. Michelle Says:

    I loaded vmfs-tools 0.20 onto FC12, and was able to access the files on my ESXi 4.1 host no problem (booting the host to ESX 4.1)

    In order to ensure my backups were good I did some md5 sums and was suprised to find problems. Not only did the source md5 not match the destination (backup), running the md5 on the same VMDK file 3 times in a row gives me a different md5 every time! Something is wrong…

    What can I do? Here’s the console output:

    SOURCE: [root@vmhost4 /]# md5sum /mnt/vmfs/FPS/FPS-flat.vmdk
    1393868fd562743fc445e5a45922a212 /mnt/vmfs/FPS/FPS-flat.vmdk [root@vmhost4 /]# md5sum /mnt/vmfs/FPS/FPS-flat.vmdk fa62dd7663fc256dd23600ced4696843 /mnt/vmfs/FPS/FPS-flat.vmdk [root@vmhost4 /]# md5sum /mnt/vmfs/FPS/FPS-flat.vmdk 190290f041844fea6f581fc8b52235d9 /mnt/vmfs/FPS/FPS-flat.vmdk [root@vmhost4 /]#

    DESTINATION(COPY): [root@vmhost4 /]# md5sum /tmp/destination/vmhost4/2011-Apr-22-rev5/fuse~unknown/FPS/FPS-flat.vmdk 190290f041844fea6f581fc8b52235d9 /tmp/destination/vmhost4/2011-Apr-22-rev5/fuse~unknown/FPS/FPS-flat.vmdk [root@vmhost4 /]# md5sum /tmp/destination/vmhost4/2011-Apr-22-rev5/fuse~unknown/FPS/FPS-flat.vmdk 190290f041844fea6f581fc8b52235d9 /tmp/destination/vmhost4/2011-Apr-22-rev5/fuse~unknown/FPS/FPS-flat.vmdk

  5. glandium Says:

    Michelle: try version 0.2.1.

  6. Michelle Says:

    Sorry – I should have said I was running 0.2.1 (I didn’t apply the -1 diff though).

    This is on a machine with ECC memory, RAID disk, etc. and no errors in the log…and seems to happen with the largest (VMDK) files I’m trying to backup.

  7. glandium Says:

    Stupid question: are these VMDKs used by a running virtual machine ?

  8. Michelle Says:

    No. The host (vSphere ESXi 4.1) is stopped and booted to Fedora 12, and then the VMFS is mounted using vmfs-tools. Nothing else running.

    Sometimes the VMDK’s copy fine and the md5sum’s match, other times not. I’ve rebooted a few times, etc. looking for clues but nothing obvious.

    That fact that I get 3 different checksums 3 times in a row suggests a code problem…unless I’m the only person with this bug…

  9. glandium Says:

    Michelle, what you could try to do is see if the following command exhibits the same problem:

      debugvmfs $device cat FPS/FPS-flat.vmdk | md5sum

    If it does (which I’d expect it to), please try running the same command under valgrind:

      valgrind debugvmfs $device cat FPS/FPS-flat.vmdk | md5sum

    and report any error it would display, preferably by email. Blog comments are not really convenient for such kind of support.

  10. Jurij Says:

    Glandium, why not to keep on the standards? I mean the mounting procedure:

    instead of: vmfs-fuse /dev/xyz /mount/point to be able use it this way: mount -t vmfs /dev/xyz /mount/point -o options

    This would be the common way to mount partitions. How long do You mean it takes until the read/write support for vmfs would be possible?

    Thanks, Jurij

  11. glandium Says:

    Jurij: mount -t vmfs would require a kernel module. VMFS-fuse is a userspace program.

    As for write support, I doubt it ever gets anywhere usable except if someone comes forward to help with it.

Leave a Reply