Another SAP backup problem

[JohnPlayerSpecial]

Hi,

Environment:

Client:AIX 5.3+NW 7.22 client+NMSAP 3.0.2
Server: Solaris9+NW7.2.2

Full Online backup fails. Only Oracle controlfile is saved to tape. I am new to AIX systems. I have never encountered this problem with Solaris boxes. I initiated this backup manually from command line as oraxyz user like this:

brbackup -d util_file_online -r /oracle/XYZ/102_64/dbs/initXYZ.utl -t online -m all -c -u system/password

Errors:

backint: cannot open /sapdataXYZ/XYZ/sapdata1/sr3_1/sr3.data1: Invalid argument
backint: error while saving /sapdataXYZ/XYZ/sapdata1/sr3_1/sr3.data1
backint: RPC error: RPC cannot encode arguments
backint: backint of backint:XYZ to backup1 failed
Session: 0, pid: 3698896 completed with exit status: 1
Unable to open output file /nsr/applogs/LGTOSAPssapXYZ.3698896. Child process 3698896 is terminated abnormally.
No output to brtools for this child.
Session: 0, pid: 3698896, total files in the session: 18, Failed. Retried 0 time(s)
Semaphore file deleted in 2 secs

 

[Rif123]

Hi,

Not that I have loads of experience of SAP, but have you double checked your .utl file so everything is set up correctly? I guess the output file might contain further information. Have you checked the permissions on the /nsr/applogs? I think you can specify another location for the baackint log in the utl-file otherwise. You should also be able to make the logging verbose.

 

[hoinz]

Hello,

what strikes me here is this:
backint: cannot open /sapdataXYZ/XYZ/sapdata1/sr3_1/sr3.data1: Invalid argument

Is this a valid file path?
Looks a bit unusual to me, not in accordance with SAP's standards, but may be valid for your system.

 

[JohnPlayerSpecial]

Path is valid. Our SAP-systems are created like that. /oracle/XYZ/sapdatas are linked to the one huge filesystem like sapdataXYZ because of easier monitoring and space allocation. This is approved by SAP.

Also I checked the permissions from similar system which is behaving normally. All the permissions are same. Backint.log owner is oraxyz:dba and it is updated normally. All the error messages I posted earlier are from backintXYZ.log

I also checked the util files but no errors there. Usually errors in util files will produce clear error messages.

 

[hoinz]

Usually errors in util files will produce clear error messages.

I don't think so. Or maybe this is not true for *.sap files.
I remember getting a very colourless error message from Legato Networker, 'cannot execute' or the like. In the end we found out that this was caused by parameter backup_root_dir in *.sap file. It wasn't used, because of backup_dev_type = util_file. But we had left it in, because of historical reasons, or maybe because of laziness.
It worked, as long as the specified dir existed. When we deleted it, backups didn't work any longer, until we deleted this parameter as well. And it took some time to find out!

So, as Rif123 suggested, double check parameter files, and perhaps delete unneeded parameters.

hope this helps

 

[emcaixlegato]

You should have linked or copied the /usr/bin/backint file to "/usr/sap/SID/SYS/exe/run/backint" Networker requires you to run their version of backint for backing up SAP on AIX. Also, you should use the nsrsapadm program to modify your nsrsapsv.cfg file in /nsr/res with the required Oracle login and encrypted password. Otherwise, permissions, etc., will be wrong.

 

[alexandre79]

Hi,

The solution is to change the Oracle parameter

filesystemio_options from SETALL to asynch

;-)