I'm running an older Dell GX280, with a new 500GB drive and 2GB RAM. Fresh install with my ~ directory copied from an older HD I got the 500GB to replace. I think I'm running ext4 on both my / and /home partitions.

And, if I run for a few days, I lose my ~/ directory. By which I mean, I can do anything with any file and directory in ~/ if I know what I'm looking for, but if I do ls, or try tab-completion, or try to view it in nautilus, it hangs. Then, I reboot, and everything is back together.

What is it? Where do I start looking for problems? I don't know where to start Googling for the answer.

Here's my /etc/fstab

jacoby@oz:/var/log$ cat /etc/fstab 
# /etc/fstab: static file system information.
# Use 'blkid -o value -s UUID' to print the universally unique identifier
# for a device; this may be used with UUID= as a more robust way to name
# devices that works even if disks are added and removed. See fstab(5).
# <file system> <mount point>   <type>  <options>       <dump>  <pass>
proc            /proc           proc    nodev,noexec,nosuid 0       0
# / was on /dev/sda1 during installation
UUID=4f677505-0b67-47b0-bbb4-858ffc1fe125 /               ext4    errors=remount-ro 0       1
# /alt was on /dev/sdb1 during installation
UUID=b0eec90c-d312-4123-b78c-7487a3347888 /alt            ext4    defaults        0       2
# /home was on /dev/sda6 during installation
UUID=1def350e-fe9a-40e3-8162-0a9f7ff8d5ef /home           ext4    defaults        0       2
# swap was on /dev/sda5 during installation
UUID=e62cd8c5-6088-44a8-84a6-7d399e42d81d none            swap    sw              0       0
/dev/fd0        /media/floppy0  auto    rw,user,noauto,exec,utf8 0       0
  • 1
    So, it doesn't go away, it causes a freeze? – lfaraone Aug 20 '10 at 2:31
  • Can you post the contents of your /etc/fstab file? – Bryce Aug 20 '10 at 4:39
  • Also, what method did you use to do the copying? And did you copy all the hidden dotfiles from your prior install? – Bryce Aug 20 '10 at 4:40
  • 1
    If there is I/O errors on the partition they should show up in /var/log/dmesg use this command to se the logfile cat /var/log/dmesg (for the current log) and cat /var/log/dmesg.0 (for the last boot-log. Please tell if there is any error messages there. Also the output from this command might be helpfull in finding the problem mount. – LassePoulsen Aug 20 '10 at 5:51
  • In addition to what Source Lab mention, one thing that may help is the output of /usr/bin/strace -s999 /bin/ls ~ when the symptoms appear. If that doesn't hang, /usr/bin/strace -s999 /bin/ls -l ~. You'll need to install the strace package first. For privacy you may want to replace some file names with xxx. Also, does this affect only your home directory, or also other directories (e.g. /bin/ls ~/bin if you have a bin subdirectory in your home)? – Gilles Aug 20 '10 at 11:20

Here's one thing I can think of that could explain your symptoms. If you have a mount point to a “flaky” filesystem, or a symbolic link to a mount point to a flaky filesystem, in your home directory, then most methods of listing your home directory could hang (including ls -l, ls -F, ls --color, but not plain /bin/ls) waiting for that flaky filesystem.

Possible examples of flaky filesystems:

  • An NFS/Samba mount where the server is not responding. This is mostly observed in unix enterprise environments.

  • A fuse mount that's hung on .

Unmounting the offending filesystem may help; umount -l or umount -f may be useful in desperate circumstances (read the mount man page before using these options). For a fuse filesystem that's waiting on a process that isn't responding, try umounting with fusermount -u (or fusermount -uz if there are open files), and killing the non-responding process.

Tip: don't mount this kind of filesystems under a directory you traverse often, such as your home directory. Have a dedicated parent directory for such mount points, e.g., ~/mnt. The same goes for symbolic links that point into these filesystems.

  • That's a maybe. I'll have to wait for the next time it happens. I AM using a CRAZY amount of FUSE. I set up my twelve mountpoints in /sshmount and symbolically link to them so that they don't show up when I du in my home directory. I can't say yet that this is definitively it, but it does seem likely. And, thanks. – Dave Jacoby Aug 20 '10 at 12:27
  • Seems like the good way to force the error is to reboot my home machine while still sshfs-mounted. I might try that later today. – Dave Jacoby Aug 20 '10 at 12:31
  • @VarLogRant: Doh, right, I should have thought of this. A crashed fuse daemon would be a modern equivalent of an absent NFS server. – Gilles Aug 20 '10 at 14:53
  • No biggie. I didn't think to mention it until you mentioned NFS. – Dave Jacoby Aug 20 '10 at 19:39
  • It happened again. I'm sure it's Fuse now. – Dave Jacoby Aug 30 '10 at 23:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.