Which level of Connect Agent are you using right now, 2.1.6 or 2.1.7 ?  We have
just releaseed a new ptf, 2.1.7.  You might want to get this new ptf downloaded
from index.storsys.ibm.com as anonymous under  /adsm/fixes/agents/aixorc.  I
think, the try-buy package you installed is still at 2.1.6 level.


I don't think this is a memory leak problem.  We have run the agent, 2.1.6,
with 500+GB of data and had no problem.  Was it your very first time backing up
the database with this EBU level 2.1.0.1.2?  If it was, were you able to start
another backup?

Another possibility is the commtimeout value set to a low value in your
dsmserv.opt file.  How full your database is?  Even it is 40GB in size, for
example, your database might be only 50% full.  When EBU was busy reading null
blocks, the backup sessions were idle for a while.  Once this idle time was
over commtimeout value, your backup sessions are cancelled by the ADSM server.
When EBU was ready to send some data to ADSM, the sessions were already
cancelled.  In this case, you should see -50, tcpip failure, in your
dsierror.log.  Since you don't see this in dsierror.log, I am curious to know
the answers to my questions above.

regards,
Thiha

>Hi,

>I recently installed Oracle ADSMConnect Agent for AIX (try-buy) on an F50
>running AIX 4.2.1 and Oracle 7.3.3 (EBU V 2.1.0.1.2)
>My obackup currently runs about halfway through until encountering the
>error: (from obackup trace file)

>EBU-4003: Process pid 78454 exited; exit code = 5
>  on 06/30/1998 20:46:55 [ 77424 : brio ]
>Process terminated with SIGNAL 4

>Signal 4 means illegal instruction which sounds like a bug.  As can be seen
>below, the obackup processes seem to be experiencing a memory leak.

>Nothing is logged in the dsierror.log file.

>The total size of all datafiles is about 40GB and the backup failes after
>about 20GB so the database size can hardly be a problem.

>Any help would be greatly appreciated.

>Vilmundur Palmason
>Systems Engineer
>Nyherji HF
>Reykjavik
>Iceland