xfs_dump problem

[Stephan L Jansen <jansen@xxxxxxxxxxxxxx>]

Hi,

We are seeing problems with xfsdump hanging in "D" state on a 1.3TB 
filesystem.
This hang seems to happen at more or less random points in the dump.   
I ran
xfs_check on the filesystem and it did not find problems.  This 
filesystem was
most likely created under RedHat 8.0 with XFS 1.3.0. This happened on 
SuSE 9.0
with kernel version 2.4.21-215-smp4G and xfsdump version 2.2.13-33 and 
now on
SuSE 9.1 kernel version 2.6.4-54.5-smp and xfsdump version 2.2.16-30.   
Once
the xfsdump gets hung running strace or gdb on the process just hangs 
so I can't
tell where the process is getting stuck.

Has anyone seen this?   Any ideas on a fix or any suggestions on 
further debugging
this issue?

There are a fair number of  files on this filesystem, here's the  
output from
"df -i":

df -i /dev/sda1
Filesystem            Inodes   IUsed   IFree IUse% Mounted on
/dev/sda1            693847296 10842420 683004876    2% /data


Here's the output from xfs_info on this partition:

xfs_info /dev/sda1
meta-data=/data        isize=512    agcount=331, agsize=1048576 blks
         =                       sectsz=512
data     =                       bsize=4096   blocks=346923667, 
imaxpct=25
         =                       sunit=0      swidth=0 blks, unwritten=1
naming   =version 2              bsize=4096
log      =internal               bsize=4096   blocks=32768, version=1
         =                       sectsz=512   sunit=0 blks
realtime =none                   extsz=65536  blocks=0, rtextents=0


Here's the command line I used:

/usr/sbin/xfsdump -l 0 -T -e -d 2048 -o -L glimpse10 -Y 7 -f 
guest@tapehost:/dev/nst0 /dev/sda1

(I know that "-Y 7" doesn't do anything under Linux, it's just a 
holdover
from my old IRIX scripts)

--

----- Stephan