can't see /

[GoogleCodeExporter]

What steps will reproduce the problem?
1.
insert 2x2Tbyte new HD on a new DNS-323 (Rev C1, updated 1.09 firmware,)
first format with vendor (separate discs, ext 2) and put alt-f
2.first start on 0.5B6
try to create a Raid 1 of 750G with sda2 and sdb2 and a raid 0 of ..anyting 
else on sda3 and sdb3, other partions not touched
3.after a lot of time resync raid 1 (2 times)
4.near 70% of 2d resync go to setettings and clic on save setting
5. later  75% go to settings and clic on save settings

What is the expected output? What do you see instead?

save settin ok 
but

mkdir: cannot create directory '/': Stale NFS file handle mount: mounting 
/dev/mtdblock0 on /tmp/mtd failed: No such file or directory 
/usr/sbin/loadsave_settings: cd: line 327: can't cd to /tmp/mtd umount: can't 
umount /tmp/mtd: No such file or directory rmdir: '/tmp/mtd': No such file or 
directory 

on ssh:

# cd /
-sh: cd: can't cd to /
# ls / 
ls: /: Stale NFS file handle

What version of the product are you using? On what operating system?

Please provide any additional information below.

Original issue reported on code.google.com by Marcelo....@gmail.com on 15 Jan 2011 at 10:36

[GoogleCodeExporter]

The reported setup and procedure is too vague.

-Was you using 0.1B6 or 0.1B6+ from the experimental release site? (0.1B6 has 
an issue when partitioning disks bigger than 1GB).

-Have you used the "advanced" partitioning, or the just the basic?

-Did you created a swap partition?

-Can you supply the disks partition table? type the following command
  sfdisk -luS

-The problem had to be transient and obviously related to NFS -- what is your 
NFS settings? Are you exporting only or mounting remote filesystem also? post 
/etc/fstab and /etc/exports

-Have you reboot meanwhile? If you are having issues with "/" and NFS the 
reboot command probably will fail (guessing) and you will have to pull the 
power cord.

Original comment by whoami.j...@gmail.com on 16 Jan 2011 at 12:21

[GoogleCodeExporter]

It happened to me just now with 0.1B6+.

-After pressing the "uninstal all" button in the Alt-F package manager, the 
login period (1 hour) expired
-I had to login again
-after login the same Alt-F package manager page appeared and I hit the 
"uninstal all" button again.
-An error window appeared saying that /Alt-F/etc/password /Alt-F/etc/... was 
not available, or something similar
-At the command line I can't see "/":
   # ls /
     ls: /: Stale NFS file handle
   # ls /etc
     ls: /etc: No such file or directory
   # cat /sys/fs/aufs/si_41909cf1/*
     /mnt/md0/Alt-F\040(deleted)=rw
     /rootmnt/root=rw
     /rootmnt/rootsq=rr
     /rootmnt/root/.aufs.xino

-NFS is not running.
-The /mnt/md0/Alt-F overlay/unioun directory was effectively deleted, abd aufs 
can't handle the situation (not its fault)
-The system in unusable

The issue is related to aufs (another union filesystem)

Alt-F packages are installed on directory Alt-F at the filesystem root chosen 
by the user at Alt-F package installation time.
A link is created at the root "/" pointing to that directory, and all files on 
that directory overlay the ones existent in "/".
See the http://code.google.com/p/alt-f/wiki/HowToFixOrCustomizeFirmware wiki 
entry
At aufs mount/umount time a set of files has to be copied from /etc/ to 
/Alt-F/etc and vice-versa, which is done by the aufs.sh script; then the 
overlay directory is unmounted.
If processes from packages are running, the overlay/unioun directory can't be 
unmounted, leaving some indispensable files in /etc in an inconsistent state.

aufs.sh needs more error checking and recovery.

Original comment by whoami.j...@gmail.com on 17 Jan 2011 at 3:28

• Changed state: Accepted

[GoogleCodeExporter]

fixed in svn now. closing

Original comment by whoami.j...@gmail.com on 31 Jan 2011 at 7:29

• Changed state: Fixed