>Othmar Weber,
>
>We are facing a similar problem. We are attempting to reduce our database
>as well.  I tried running the DELETE DBvol command and it didn't work.
>I contacted IBM and they said the same thing, run DUMPDB and LOADDB.
>It sounds like you have run the dump and load commands and have experience
>some problems...yes/no?  If so, what were they and how did you resolve
>them.
>
>
>Thanks
>Vince Kelley

So it seems to be of common interest: Here are my experiences using the
DUMPDB/LOADDB/AUDITDB commands.

My first contact was in version 1 of ADSM. Doing recovery tests
i tried to use these commands. Dumping the Database (>1GB) lasted
about one hour but reloading whould have lasted 2 days so i stopped the
test (CPU time is expensive during the week). I was hoping for an APAR
which promised better performance. Before I could make another test I
migrated to version 2 of ADSM. Because of the new BAckup DB command
there was no need for further testing the DUMPDB/LOADDB. Our server
did well until it crashed 3 weeks ago (good luck it was friday). This
is what happend:

A BAckup DB process did not start correctly. After that the recovery
log started growing (LOGMODE=NORMAL). After two days the LOG was at
90%. I tried to cancel the BA DB process but that did not work. Then
I stopped the server. During the restart phase it consumed about one
hour of CPU time and then it got a memory allocation error (a region
size of 40M seems to be not enough in such a situation). I increased
it to 100M and tried to restart the server again. One CPU hour later
it went to a wait state and did nothing anymore. No CPU time was used,
no IO was done, no paging; just nothing. I opened a severity one problem
with IBM and the guy there told me to install a new server and rescue
the old database using the DUMPDB/LOADDB/AUDITDB commands. The size
of the database was >2GB and the size of the LOG was about 1GB at the
time of failure. The DUMPDB command lasted about one hour. The reload
command about 20 hours. Saturday evening I started the AUDITDB
command which I canceled at monday evening when about one third of the
database entries was reloaded. The JOB consumed about one CPU day.

In the meantime I installed a third ADSM server using the last database
backup which was three days old (remember the database backup process
did hang). This was done in three hours. Then I started auditing all
Volumes which lasted about 1.5 days.

That was the longest weekend of my life.

Thats why I don't want to use the DUMPDB command if there is another







































































possibility.  If everything with the database is fine and the server was
quiesced before halting no AUDITDB would be required after LOADDB. But
which ADSM administrator knows every bit of 'his' database?
Just for info: My server is at level 0.9.

Good Luck!

Othmar Weber
Bayer AG Germany