Hi all!
I haven't seen anyone on this list mention a problem like this before so
I thought I would post it to see what everyone thinks.
First, our configuration:
Server:
HP T520 running HP-UX 10.20
ADSM version 2 release 6 level 0.12
2 BreeceHill DLT Libraries w/ 4 DLT4000 drives each
Clients:
various HP-UX ADSM clients
The problem:
We run backups of our production environment to one DLT library and
non-production to the other DLT library. We further segregate our
backups by running two classes of backups- system and database. Of
course, based upon a schedule, the backups might be true backups or they
might be archives... We have some pretty elaborate scripts that automate
what to backup, how to back it up, etc.
All of this works well, and has been doing a pretty reliable job for
some time now- until recently. We have recently started experiencing a
frequently recurring problem where our client will begin a dsmc process
to back up a specified filesystem or virtualmountpoint and from that
point on it does nothing but hang, repeating a message stating that it
cannot find the real file mount point for a particular file system or
directory, other than the directory to be backed up.
Example: assume we run a dsmc incremental -node=<node>
-pass=<pass> -subdir=yes /opt/data1
Then we get a message like the following:
Then we get a message like the following:
Couldn't find real mount point for filespace /u02/applmgr/product/10.5
This becomes a problem first because the directory or filesystem that is
being backed up isn't the directory that is stated in the error message.
It should not even be looking at the directory or filesystem because it
is not in an part of the directory tree being backed up.
Secondly, we capture error messages returned at run time so that we have
additional information besides the often sparse server log. As you can
imagine, this becomes a problem since this error message repeats rapidly
for hours and hours consuming up to 1.5 GB until an operator gets a disk
space warning message and can alert someone to look at the problem.
This problem happened once a few weeks back- but didn't occur regularly
and we were unable to reproduce it. But now it has begun to happen about
two to three times a week. It has happened 4 times in the last 10 days.
We are not sure if there is a known bug in the ADSM code for HPUX, but
we are unsure why this configuration has worked without much incident
for quite a long period of time.
Has anyone heard of such a problem or experienced a similar situation? I
can give more elaborate details of our procedures and backup methods if
it will help. We are in the process of logging a support call but wanted
to try this avenue as a possible means of learning additional
information if it is known.
Thanks in Advance,
Robert Alexander
MCI Systemhouse
|
