Backup under Cluster env.

[santam2007]

We have difficulty on creating backup saveset for MS Cluster.

NWS: v7.3.2
NMSQL: 5.0

Here are the details:
CISUQ198 and CISUQ199 is a pair Cluster Group for SAP UQ system. Within this Cluster Group, CISUQ199 is the primary node for DB Instance, whereas, CISUQ198 is the primart node for SAP CI Instance. Under normal circumstance, these two servers are in an "active/active" status, however, when failover, the remaining server will take up both task of DB and SAP CI Instances. Each instance has its own SAN Disk resources list below and SAN Disk R: and P: are normally attached to server CISUQ198. Given this is the design for all SAP Cluster Servers on QA and Production environment.

DB Instance (CISUQ199)
1. Disk O:
2. Disk T:
3. Disk U:
4. Disk V:
5. Disk W:

SAP CI Instance (CISUQ198)
1. Disk P:
2. Disk R:

For DB Instance, we use alias CISUQ102
SAP CI Instance, we use alias CISUQ103
and the MS Cluster named CISUQ101

How can we configure the saveset to backup (i)all Disk drives (file level) of both Instance, (ii)MSSQL under DB Instance and (iii)local C: drive of both physical node ?

To backup "O,T,U,V,W" should I config saveset with client name "cisuq102" ?
For "P,R" should I config saveset with client name "cisuq103" ?
For "MSSQL backup" should I config saveset with client name "cisuq102" and command "nsrsqlsv -a cisuq102" ?
For "local C:" should I config saveset with "cisuq199.domain.com" ?

Thanks

 

[haasemi]

Hi,

for the physical cluster members (like CISUQ199) you should create a client with saveset "ALL". This will backup the local drives and the system savesets of the physical node, but not the drives which are owned by the virtual ckuster nodes.

The drives of the virtual nodes (like CISUQ102) should be saved with his owne client definiation/name.
E.g. CISUQ102 with drive O:, T:, ...
and CISUQ103 with P:, R:.
It could be possible that you have to use the additional parameter -a <virtual node name> for the save/savegrp command, but often it work without.
Also it could be possible that you have to use the "ignorepathownership" file, if the pathownerships are not set right.

Your idea to backup the SQL DB is right (corresponding to my explanation before).

Don´t forget a client for your "Quota Disk" of the cluster.

regards
Michael

 

[santam2007]

Thanks, I did it