df and zettabytes

Posted on June 23, 2009 by ward

This is a very confused filesystem. But check it out – df supports zettabytes!

# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/md0               19G  -64Z   22G 101% /

Surprisingly, the machine is up and runs just fine.

This entry was posted in Sysadmin and tagged , , , , . Bookmark the permalink.

4 Responses to df and zettabytes

  1. emailtoid.net/i/eca7e350/… says:
    June 23, 2009 at 11:06 am

    It would be intersting to see some more:
    fdisk -l
    mount
    …

    What is causing this ? Sparse files ?

  2. ward says:
    June 28, 2009 at 3:59 pm

    Well, there’s nothing special in mount or fdisk -l.

    /dev/md0 on / type ext3 (rw,errors=remount-ro)

    It may be caused by sparse files. This is a xen dom0 that was used for some iscsi experimentation a while back and this may be a result of that. But I don’t know exactly what the person doing the iscsi testing did, so it’s hard to say.

    It’s an old 2.6.18-xen kernel. But I have loads of those running still, and this is the only box I’m seeing this on. The machine’s been up for a long time – 607 days currently. This is a production machine, and du -sh is normal:

    # du -sh
8.5G	.

    So I’m not too worried.

  3. wo.uter.be/ says:
    July 3, 2009 at 2:30 pm

    This kind of thing usually points to corruption.

    If I were you, I’d take the filesystem offline immediately, and run ‘fsck -f’. If that finds nothing, no worries — but better safe than sorry.

    It’s quite unlikely that this is related to sparse files. Really.

  4. ward says:
    July 3, 2009 at 2:52 pm

    Yeah, fixing this is obviously on my todo list. But like I said, this is just a dom0 filesystem on one system in a cluster, and the domUs are in lvm slices. The domUs are important; the dom0 as such is not very important.

Leave a Reply