Dustin J. Mitchell wrote:
On Wed, Nov 26, 2008 at 10:55 AM, Jim Summers
<[EMAIL PROTECTED]> wrote:
Looking on the clients, I found no activity except on one gtar process on
one DLE. That client also has amdumps and sendsize running. Using very
little cpu.
I think that this is the key to what stalled the backups. Was the
gtar process hung too? Such a hang can be caused by a hung mountpoint
or, in some fs's, filesystem corruption. But Amanda will dutifully
wait for more data.
Killing this process and re-running amcleanup should get you a working
setup again, but I assume you've figured this out -- it's been a few
days since you posted..
Apologies for letting this go so long, but I wanted to see if the changes I
made isolated the problem.
I decided to temporarily remove the DLE that was being processed on the client
when things began hanging. That did seem to make a difference. The nightly
backups began operating as expected again.
I am going to start perusing through that DLE and see if I can see any
problems. The DLE is pointing at a "shared" directory that many researchers
use. Is there any chance that symbolic links could be causing a problem?
There are also directories within the DLE that are being excluded by the
amanda exclude file. I was thinking that is might be a problem if a link
pointed across those excludes?
I will try to report back when I discover something.
Many Thanks
Dustin
--
Jim Summers
School of Computer Science-University of Oklahoma
-------------------------------------------------
