Start a Conversation

Unsolved

This post is more than 5 years old

1178

September 25th, 2006 11:00

SAP /R3 Cluster

Hello all,

We have the following enviorenment:

Physical nodes:
escorpssappc
AIX 5.3
NetWorker 7.2.2
SAP R/3 4.6c
Oracle 9.2.1
NM for SAP 3.02

escorpssappd
AIX 5.3
NetWorker 7.2.2
SAP R/3 4.6c
Oracle 9.2.1
NM for SAP 3.02

Virtual Nodes:
pkgsap
Instance SAP/R3

pkgoracle
Instance Oracle

*The virtual nodes are normally on different physical nodes.

How do we configure NetWorker to run backup of SAP/R3?, do we have to create a client for pkgsap or for pkgoracle?

14.3K Posts

September 25th, 2006 11:00

Yes you do.

14.3K Posts

September 25th, 2006 12:00

You will need to define client resource for physical and virtual clients and SAP data which I assume belongs to virtual node should be backed up via virtual name. Thus it doesn't matter on which physical hosts it run. I remember having DB2 in AIX cluster as storage node and it did take some time to figure out how to configure it due to nature of HACMP. In this case as far as I see there is no storage node component so that makes it easier to configure.

253 Posts

September 25th, 2006 12:00

Sorry Hrvoje, I don't understand you. Can you explain anymore?

Thanks in advance.

253 Posts

September 25th, 2006 12:00

No, they are Dedicated Storage Nodes. But we needed to know if we must create the client 'pkgoracle' or 'pkgsap'. Which one of the virtual resources do we have to include at the field Save set the clause backint:oracle_instance?

14.3K Posts

September 25th, 2006 13:00

Not supported - been there :D Actually you may wish to force it to write index on different clientname, but at the end officially support may turn you down once you get in some sort of problem.

So, to have it supported you will need to make it storage node instead of dedicated one. The rest is configuration as manual states. You may wish to dicuss this with support too - especially if you are not end customer but doing it for someone else (imagine what happens when end customer finds out that it has unsupported solution).

14.3K Posts

September 26th, 2006 08:00

Because EMC says that physical hosts should backup data belonginh to physical hosts and virtual one should backup to virtual ones, right? Do you see the conflict now?

253 Posts

September 26th, 2006 08:00

Sorry, the enviorenment is:

Physical nodes:
escorpssappc
AIX 5.3
NetWorker 7.2.2
NM for SAP 3.02

/dev/hd4 0.12 0.11 12% 2221 8% /

/dev/hd2 1.75 0.16 91% 39591 49% /usr

/dev/hd9var 0.12 0.07 46% 668 4% /var

/dev/hd3 0.75 0.08 90% 581 3% /tmp

/dev/hd1 0.12 0.12 1% 13 1% /home

/proc - - - - - /proc

/dev/hd10opt 0.12 0.07 41% 1023 6% /opt


escorpssappd
AIX 5.3
NetWorker 7.2.2
NM for SAP 3.02

/dev/hd4 0.12 0.11 12% 2221 8% /

/dev/hd2 1.75 0.16 91% 39591 49% /usr

/dev/hd9var 0.12 0.07 46% 668 4% /var

/dev/hd3 0.75 0.08 90% 581 3% /tmp

/dev/hd1 0.12 0.12 1% 13 1% /home

/proc - - - - - /proc

/dev/hd10opt 0.12 0.07 41% 1023 6% /opt


Virtual Nodes:
pkgsap
Instance SAP/R3
SAP R/3 4.6c

# lsvg -l P53_CIvg
P53_CIvg:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT
sapmnt_vol jfs2 30 30 1 open/syncd /sapmnt

sapP53_vol jfs2 30 30 1 open/syncd /usr/sap/P53

saptrans_vol jfs2 249 249 1 open/syncd /usr/sap/trans

P53adm_vol jfs2 15 15 1 open/syncd /home/p53adm

sapinstall_vol jfs2 259 259 1 open/syncd /SAPinstall

Loglv01 jfs2 1 1 1 closed/syncd N/A

loglv00 jfs2log 1 1 1 open/syncd N/A

deofastlv jfs2 22 22 1 open/syncd /opt/DEOFAST

pkgoracle
Instance Oracle
Oracle 9.2.1

# lsvg -l P53_ORAvg
P53_ORAvg:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT

oracle_vol jfs2 7 7 1 open/syncd /oracle

oracleP53_vol jfs2 188 188 1 open/syncd /oracle/P53

stage_vol jfs2 108 108 1 open/syncd /oracle/stage

client_vol jfs2 6 6 1 open/syncd /oracle/client

sapreorg_vol jfs2 119 119 1 closed/syncd N/A

sapcheck_vol jfs2 31 31 1 open/syncd /oracle/P53/sapcheck

templv jfs2 4 4 1 open/syncd /oracle/P53/sapbackup

loglv02 jfs2 1 1 1 closed/syncd N/A

loglv01 jfs2log 1 1 1 open/syncd N/A

/oracle/P53/sapdata1

/oracle/P53/sapdata2

/oracle/P53/sapdata3

/oracle/P53/sapdata4

/oracle/P53/sapdata5

/oracle/P53/sapdata6

This environment is supported?

We know that to run SAP's backup, we have to configure the instance 'pkgoracle' in NetWorker . Now then, the problem comes because the SAPMNT isn't shared and from 'pkgoracle' is not access to the brtools (SAPMNT). We think that it is the real issue.

Thanks and regards.

253 Posts

September 26th, 2006 08:00

I don't understand why the solution is not supported. Can you explain me?

Thanks very much.

14.3K Posts

September 26th, 2006 08:00

This environment is supported?

What I said before was that not supported is to have cluster solution and dedicated storage node.

We know that to run SAP's backup, we have to
configure the instance 'pkgoracle' in NetWorker . Now
then, the problem comes because the SAPMNT isn't
shared and from 'pkgoracle' is not access to the
brtools (SAPMNT). We think that it is the real
issue.

I see what you mean. Is it SAP design fot SAPMNT not to be shared in cluster configuration?

253 Posts

September 26th, 2006 08:00

Is it SAP design fot SAPMNT not to be shared in cluster configuration?
We think so. But we think that it is necessary to have accessible from 'pkgoracle' the brtools which are in /sapmnt and /usr/sap/P53. Are you OK?

We have thought to create a folder in both physical nodes with the content of /sapmnt. This way, in the variable SAP_BIN of the file .cfg indicates the path of the created folder. What do you think?

Thanks very much

14.3K Posts

September 26th, 2006 09:00

I believe SAP should answer that question. Legato tends to call brtoold so if brarchive or brbackup can't get to data it requires it seems something is wrong with how it is configured in cluster. Usually SAP people are very touchy about those details and they should know the answer.

I guess what you suggest may work, but SAP would need to give green light that it approves it. Even then a restore test should be performed to verify it. If there would be supported and "no workaround required" approach I would first followed that up. Try to talk to SAP people as they should know it for sure.

253 Posts

September 27th, 2006 04:00

SAP has given us the solution, the Cluster does not work fine and is going to be re-formed. The new configuration is going to include the /sapmnt shared for both virtual nodes.

Thank you very much for your help.
No Events found!

Top