cancel
Showing results for 
Search instead for 
Did you mean: 

UNIX AND NETBACKUP

891074300
Level 3
Accredited
hi
can anyone give me a guideline on steps to follow when installing the SAP agent on any unix platform? I am particularly interested in knowing what scripts to edit..last time I tried to do the installation using the installation guide...and it turned out to be a big flop.

the situation is as follows, I have a windows master server on Netbackup 6.0 MP4 and two unix clusters running SAP...the unix servers are running on Tru64 and the other on HP-UX.

kindly assist.

regards

JW
1 REPLY 1

Anonymous
Not applicable
It might be best to start from what is working and go from there. So on Tru64 cluster...

Firstly, can you perform a file system backup of your Trucluster.?
If not what happens, if so good....

Assume at the least it is a 2 node cluster, and you run the database instance on one of those nodes, and if one becomes unavailable the database instance fails over to other node - this is all controlled via CAA within Tru64.
Likewise you should have netbackup setup in this way also.

To get your status of cluster applications type:
# caa_status -t

This give a summary of where your caa controlled apps are running within the cluster and there status.

When you install the DB agent for SAP it will also place script template files in
/usr/openv/netbackup/ext/db_ext/sap/scripts/sap_oracle
/usr/openv/netbackup/ext/db_ext/sap/scripts/sap_maxdb

Use the scripts for your SAP db and take a copy and edit those.
Files to edit or pay close attention are
initSAP.sap
initSAP.utl
You might want to get SAP Basis person involved if you are not sure about stuff in these files - I know I did - it helped a lot. And go back and read the SAP Agent Admin Install guide again, it is helpful.

Other files to edit are those that Netbackup calls from its policies.
sap_online_backup # this will be script to call from the policy
sap_offline_backup # ditto
sap_redolog_backup # archive logs script


You will need to edit at least the above files and replace most entries or lines that contain orasap to be ora
being the SAP Instance identifier.

Finally in our Trucluster environment we added a Start script that Netbackup runs instead of the above _backup files, however this Start script does call the _backup file eventually, only after it has found where the SAP DB is running within the Cluster. ie using the output from caa_stat. As we perform the backup local to the node the SAP DB is running on to help performance.
With this we also try and run Netbackup on the same node as the SAP DB.
Assuming you only have one instance on this cluster.

For a bolts and braces approach we also do the following.
Lastly, we edited the backup_exit_notify script on the master, and also created a backup_exit_sap script

backup_exit_notify is run at the end of every Netbackup backup.
We have it check for a policy of the SAP type. If so then call backup_exit_sap.

We have a naming convention for our SAP policies, and the _backup_exit_sap file checks if the policy name just run meets the criteria of a Database backup, then it follows that a redolog backup should be performed immediately afterwards. Following on if the preceeding backup happened to be an redolog backup we perform a filesystem backup of some SAP specific config files - as we refer to them - Profiles.

This is our backup workflow, and was born back in the NB 3.4 days by a Veritas consultant and we have tweaked and updated since. It may work for you, may not but is an outline of where to look and develop your files.