SAP APO Livecache Multiple instance and MC-ServiceGuard

[Andrew E. Malone_1]

We have planned a MC-ServiceGuard 11.14 Cluster for SAP APO with SGeSAP B.03.08. The production instance of livecache is planned to fail over to a QA system also running LiveCache and the complete APO database. The problem is the existence of two lc admin accounts: lcaadm and lcqadm on the failover node. The instructions in the SGeSAP doesn't address the interaction between the ownership of /sapdb/programs. Any insights would be greatly appreciated.
Thanks in advance.

[Geoff Wild · Accepted solution]

We arn't using APO - but we do have SAP Prod and QA in our MS/SG environemnt.

What we do is check for QA in our Prod control script:

function customer_defined_halt_qa
{
print "Checking if qapkg is running"

STATE=`cmviewcl | grep iqa | awk '{ print $2 }'`
if [[ $STATE = "down" ]]
then
echo "QA is down...starting up IPR"
else
if [[ $STATE = "up" ]]
then
QA=`cmviewcl | grep qa | awk '{ print $5 }'`
print "QA is running...on the current host: \c"
print ${QA}
if [[ ${QA} = $(/usr/bin/hostname) ]]
then
print "halting qapkg on ${QA}"
/usr/sbin/cmhaltpkg qapkg
fi
fi
fi
}



Then in the main program area:

# Test to see if we are being called to run the package, or halt the package.

if [[ $1 = "start" ]]
then
print "\n\t########### Node \"$(hostname)\": Starting package at $(date) ###########"


verify_physical_data_replication # add hook for MetroCluster

customer_defined_halt_qa

activate_volume_group

activate_disk_group

customer_defined_fsck_mount

check_and_mount

verify_ha_nfs $1 # add hook for NFS

add_ip_address

get_ownership_dtc

customer_defined_run_cmds

start_services

start_resources

Rgds...Geoff

[Andrew E. Malone_1]

Thanks Geoff,

The critical point I missed in the initial question was that the QA Livecache is not going to be shut down during the fail over hence the problem.

Andy

[Geoff Wild]

Ok - then you have 2 choices (maybe more):

1) ensure the environments are completly seperate - IE: both don't use /oracle....

If they both need their own /usr/sap/trans or /APPL - then you may be in trouble...

2) add a third node to the cluster and run QA on that one - and never failover Prod to it....

Rgds...Geoff

[Andrew E. Malone_1]

Thanks again, Geoff but the SAP environment is fixed in the firmament with QA as the failover. It works without a hitch on other SAP configurations onsite and We are a good sized SAP environment. I am just trawling for some other experience and insights into APO which is new to use for stablity and appropriateness.

Regards,

Andy

[Volker Borowski]

Hi Andrew,

never had APO in my hands, but a few standard-SAPs with SAPDB. Don't they have
seperate Mountpoints for each SID in APO ??

/sapdb/QAS
/sapdb/PRD

So you should be able to use two filesystems with diffrent owners and versions. Just make sure that the numeric ID of the Users are diffrent !

You might need some tricks for the TCP-IP-Server (vserver/x_server) of the database. Not sure, how the port is changed, or if you can run it anyway, even if the PRD DB is not failed over.

As far as SAP is concerned, you need diffrent SAP Instancenumbers between Q and P, so that the SAP dispatcher-port is not blocked when you switch.

Hope this helps
Volker