ECC on cluster

[Former Member]

Dear Gurus,

Need your guidence to understand doubts regarding SAP on cluster configuration.
we are running ECC 6.0 on AIX 5.3 and Oracle 10g.We are using HACMP Clustering solution to implement Active Passive cluster configuration.

ASCS +CI+DB are running on primary node.And secondry node usually remains Idle.

On primary node we are having below mount points

/
/usr
/var
/tmp
/home
/proc
/opt
/usr/sap/trans
/oracle
/sapmnt/SID
/usr/sap/SID
/oracle/SID/102_64
/oracle/SID/mirrlogA
/oracle/SID/mirrlogB
/oracle/SID/oraarch
/oracle/SID/origlogA
/oracle/SID/origlogB
/oracle/SID/sapdata1
/oracle/SID/sapdata2
/oracle/SID/sapdata3
/oracle/SID/sapdata4
/oracle/SID/sapreorg
/oracle/stage/102_64

On secondry node we have only below mount points

/
/usr
/var
/tmp
/home
/proc
/opt

When we tried switchover we found User's SAP sessions are getting terminated and SAP related mount points from Primary node is mounted to secondry node and then SAP is started on secondry node. Now client requirement is when switchover takes places there should not be a downtime and no user's unsaved work shuould be lost.

Can i have your opinion on the same is this normal process in active passive clluster configuration that users sessions terminate during Switchover/Failover ?

If not what i need to check in my landscape to rectify this.
If yes how can we achieve above objective.

I have gone through many threads and FAQ's on http://scn.sap.com/docs/DOC-25454 but not clear on cluster side.

Regards

[Former Member]

Hi Kuldeep,

I agree with sunil that you should utilize your both nodes by running services on both nodes like 1node DB and other node ASCS+CI in this case your resources will be utilize and at the time of Switchover/Failover downtime will be minimize to sift only 1 service.

in this scenario at the time of Switchover/Failover only CI node failover will make a downtime otherwise db node should not make any downtime.

We are also using HP sgesap and having this information.

Regards

Gajanand Gupta

[Former Member]

>>ASCS +CI+DB are running on primary node.And secondry node usually remains Idle.

I suggest you should also make use of secondary node, installing dialog instance or move DB to secondary node. This will make utilization of your infrastructure and boot some performance.

>>Can i have your opinion on the same is this normal process in active passive clluster configuration that users sessions terminate during Switchover/Failover ?

There will be a downtime as the services will get re-started to another node when they switch.

go through http://scn.sap.com/docs/DOC-7848

[Former Member]

Dear Sunil

Thanks for your reply !!!!

Currently we do not have much resources available on secondry node (Even i do not know why we went for clustering !!! ). our client is planning for new hardware ,may be in another 3-4  Months and we are looking in all possiblites including the ones suggestedby you for future and for the same we are collecting information based on the practical experience of the SCN users.

if we have CI+ASCS+DB on one node and DI on another node :

1  Will you call this as active active clustering or active passive clustering ? 

2  Now in this scenario if Primary node goes down ASCS and DB will be shifted to secondry node

    Transactions at DB level will be rolled back (Since complete DB will shift and restart ) or will

    resume after 1-2 min halt ???

3  What about Transactions running at SAP level and the enqueue locks occupied by them ??  

My main concern is if ASCS  and DB shifts to or restarted at secondry node is it possible to retain users sessions ??

Regards

Kuldeep

[Former Member]

First I suggest you to check below as the OS and oracle is out of support.

>>AIX 5.3 http://www-03.ibm.com/systems/power/software/aix/v53/index.html

>>Oracle 10g - Note 1110995 - Extended maintenance for Oracle Version 10.2

>> if we have CI+ASCS+DB on one node and DI on another node :

>>  1  Will you call this as active active clustering or active passive clustering ? 

This will be active passive cluster configuration.

I suggest you to configure CI+ASCS on node A and DB on node B.

Install additional dialog instance on Database instance (node B in this scenario). When CI+ASCS move to node B then the DI will be stopped and CI+ASCS and DB will be running. (No harm if you want to run also the DI but you may run out of resources.)

BTW whats the NW version you are on? as there are some changes in NW 7.3.

Refer - http://scn.sap.com/community/netweaver-administrator/blog/2012/05/22/whats-new-in-sap-netweaver-73--... part 1 and part 2

also go through the installation guide which has the details of HA configuration but you have to discuss the HA with hardware vendor.

[Former Member]

Dear Sunil

We will go for supported  OS and DB  during hardware migration (using Export/Import Technique)
And for the same we have to reinstall SAP.

Since we observed there is downtime and interruption in user work during Failover in current cluster design so we are plnanning to change cluster design.

Even if we go to cluster design recommended by you ASCS+CI on Node 1 and DB on Node 2 :

My doubts are still same

1 Say Node 2 (Having DB)  goes down , Database File system will be shifted to Node 1 : what will happen to transactions at DB level  ( Since complete DB will shift and DB restart will take place)  ?

2 If Node 1 goes down what about transactions running at SAP level and the enqueue locks occupied by them ??  

Is there any Cluster design in which if Primary node goes down user's may do not even feel and continue their work normally.

We are on NW 7.0

Regards
Kuldeep