RMAN backup fails - NetBackup - No action No backup

Hi Folks

I have an Oracle9i instance registered with the Oracle9i RMAN repository 
instance, libraries are linked into NetBackup and the configuration parameters 
are set.

When a backup is scheduled the database is not backed up, no logs are produced, 
nothing happens, the backup profile is called, runs and does not backup.

Has anyone seen anything like this?

I have around 150 databases a mixture of 9i and 10g, which are configured in 
the same way and all function as expected, schedule a backup and it works.

Yet this problem instance does not backup.

Neither NetBackup or Oracle have anything in the knowledge bases which come 
close.

Any thoughts would be greatly appreciated.

Regards

Jim

Jim,

I use RMAN exclusively but not to NetBackup instead we write to NFS so I have 
no direct experience with what you are dealing with, but, a long shoot here, 
when you say:

> When a backup is scheduled the database is not backed up, no logs are 
> produced, nothing happens

and then you say

> ... the backup profile is called, runs and does not 
> backup.
> 

How do you know the backup profile is called if there are no logs?  What I am 
thinking is this -- do you have this in a script on a *NIX server?  Can you 
check if the script is executable?  Many times I've written a cronjob, 
scheduled it and it never ran because I'd forget to "chmod +x <script>.sh" ...

Regards,
- Vitaliy

Hi Vitaliy

The way the backup works is:

   Run script db_bkp, this sets a flag in a file called backup_request.

   This file gets polled between 22:00 and 01:00, if the flag is set to yes
   then an RMAN backup kicks in, all databases have a NetBackup policy
   associated to ensure that the backup runs to the correct tape silo.

   Scripts are centralised and are used for all database backups, so I can rule
   out permissions as other backups run to completion.

   The database in question is definitely registered with the repository
   database and configured for the tape subsystem.

   The file which gets polled is definitely readable.

I have had this executed in real time with NetBackup Admin watching the policy 
wake up and be executed and call the script to run the backup which as you know 
is fairly standard stuff.

i.e:

connect target '/'
 shutdown;
 startup mount;
 run { backup database tag STGI1_080818_22h02 format '%d_%T_COLD_%U'; }
 shutdown;
 startup;
 allocate channel for maintenance type sbt;
 delete force obsolete;

Yet, the database is never shutdown and RMAN itself never runs.

Any thoughts?

Regards

Jim

Jim,

Ok, so we know that everything up to the point of calling RMAN script is 
working fine.  What happens if you call the RMAN script manually, live?  If you 
have to strip it down to bare minimum then try something like this:

connect target '/'
 shutdown;
 startup mount;
 shutdown;
 startup;

How is the environment set when the RMAN script runs?  Is the call made via:

   su - oracle -c "/path/to/script.sh"

where it picks it up from oracle's profile or does it use some other method?

Regards,
- Vitaliy