Netweaver CE 7.1 ehpk1 upgrade to SP04 on MSCS

[Former Member]

Hi

I am building our production environment. I have installed NetWeaver CE 7.1 on a HA Cluster - using MSCS. Both app instances are running fine, can failover and back etc.

I am having issues upgrading to SP04. I already did this on our DEV and QA standalone systems without issue.

I am running this from the primary instance.

The JSPM bombs out when trying to synchronize bootstrap for the instance on the secondary instacne (host irlcvap0028). The issue I think is that it is trying to run a command on the D:\ which is local to the secondary instance, so the primary instance cannot access it via D:\usr\sap.... the sync is successful on the primary instance (host irlcvap0027). The log for SAP CONTROL, shows Permission denied.

How do you upgrade to SP04 on a MSCS?? Is it different to a standalone upgrade? I thought I could upgrade the primary instance, and then either upgrade the secondary instance, or copy the binaries. I cannot find a procedure specific for MSCS environment.

Help appreciated - go live is next month!

Thanks

Rosie

I have copied in some logs below.

JSPM_MAIN_1_01.LOG

Process ID 24, name sapcontrol has been started.

Apr 8, 2011 12:26:32 PM [Info]: Command line: D:\usr\sap\CRK\J00\exe\sapcontrol -nr 00 -host irlcvap0028 -function ParameterValue DIR_PROFILE

Apr 8, 2011 12:26:32 PM [Info]: Standard out: D:\usr\sap\CRK\J00\j2ee\JSPM\log\log_2011_04_08_12_18_14\SAPCONTROL_PARAMETERVALUE_00_05.OUT

Apr 8, 2011 12:26:32 PM [Info]: Process ID 24 has been started.

Apr 8, 2011 12:26:32 PM [Info]: Waiting for process ID 24, name sapcontrol to finish.

Apr 8, 2011 12:26:32 PM [Info]: Process ID 24, name sapcontrol has been finished, exit code 0.

Apr 8, 2011 12:26:32 PM [Info]: Process ID 25, name sapcontrol has been started.

Apr 8, 2011 12:26:32 PM [Info]: Command line: D:\usr\sap\CRK\J00\exe\sapcontrol -nr 00 -host irlcvap0028 -user IRLPROD\svc_miicluster <SecureField> -function OSExecute ""$Start_Program_01"" 0 300 sapcontrol_rmi_sync_bootstrap.log

Apr 8, 2011 12:26:32 PM [Info]: Standard out: D:\usr\sap\CRK\J00\j2ee\JSPM\log\log_2011_04_08_12_18_14\SAPCONTROL_OSEXECUTE_00_01.OUT

Apr 8, 2011 12:26:32 PM [Info]: Process ID 25 has been started.

Apr 8, 2011 12:26:32 PM [Info]: Waiting for process ID 25, name sapcontrol to finish.

Apr 8, 2011 12:26:32 PM [Info]: Process ID 25, name sapcontrol has been finished, exit code 1.

Apr 8, 2011 12:26:32 PM [Error]: Unable to synchronize bootstrap for instance J00. The possible reasons: Kernel is older than expected.

SAPCONTROL_OSEXECUTE_00_04.OUT OSExecute

FAIL: Permission denied

[Former Member]

The solution for the problem (Windows):

Remove SAPService<SID> from the administrators group on all the secondary servers.

Make sure the user is not a part of group that is in administrators group.

set parameters on default instance:

service/protectedwebmethods=DEFAULT

service/admin_users = <sid>adm

restart sapstartsrv service on all secondary servers.

[Former Member]

Thanks Moshe

Your solution helped me to resolve the similar issue I was facing in updating 7.4 through SUM tool.

Thanks

[Former Member]

Thanks Sunny.

I did as per the note - and took the 2nd clusterd server off-line.

But I get the same behaviour. Do I need to move out the profiles relating to the the 2nd clusterd instance temporarily, so the 1st cluster node doesnt know about it ? These profiles are in the shared globabl disk?

Is there anyone who has upgraded to a NW Service Pack in a MSCS HA environment.?

[sunny_pahuja2]

Hi,

Yes, you need to stop 2nd node so that upgrade tool should not know about this.

But I have a question here regarding your HA config, if your primary node is down then second will be up. But in this case how you will run visual admin, sdm, JSPM etc. because these reside on the first node locally ?

Thanks

sunny

[Former Member]

Yes - I had stopped the 2nd node.

I have JSPM locally on both nodes. I installed two application instances for HA as per the install doc.

I think the reason that JSPM is picking up the 2nd node, is from the profile of the 2nd instance which, going by the logs is read first.

Should I try moving the profile for the 2nd node to a tmp directory while upgrading the first node?

How do you normally upgrade the 2nd node then?

Thanks

Rosie

[sunny_pahuja2]

Hi,

I doubt this cluster installation method for As Java system because now you have 2 separate JSPM instances. How you will upgrade both ? If you try to manually upgrade on node 2 as well then it will create problem because 2nd node does not have database. Please raise an OSS call with SAP.

That is the reason I never install AS Java system in cluster. Because when failover will happen then 2nd node will restart and since server nodes of java takes time to restart,so, it is like downtime of the system which defeat the purpose of HA. Anyway raise an OSS call with SAP.

Thanks

Sunny

[Former Member]

Thanks Sunny.

I have OSS note open already. No progress yet.

Would you recommend I design a different HA/DR strategy? This was the recommended by SAP.... I understand in a failover secanario Java will need to start up on the 2nd node, so it will not be automatic. But as a business we can live with this.

Technically I should be able to upgrade? Or is it after this install you can never upgrade? I followed the SAP HA install guide for NW CE 7.1.

What is the recommended HA strategy for an AS Java system.?

Thanks for your response.

[sunny_pahuja2]

Hi,

Ya, it is SAP supported scenario but my personnel opinion is don't install As Java on HA because when failover happens that time Java takes time to restart which counts as downtime of the system. I prefer As Java DR then HA solution.

I am not sure how you will upgrade now. Let see what SAP reply.

Thanks

Sunny

[Former Member]

hi Rosie,

How did you solved the problem eventually?

[Former Member]

One additional piece of information here.

I did upgrade the JSPM, the J2EE KERNAL(7.11.20 to 7.11.61.0) & the SAP JVM (5.01.24.0 - 5.01.45.0) & the BC-FES-IGS (7.10.5.0 - 7.10.8.0). These components deployed successfully on the primary instance.

Is it possible I need to copy the relevant binaries manually to the secondary instance - before the rest of the components will upgrade to SP04?

The other part of the error is related to the Kernel.

Apr 8, 2011 11:54:32 AM [Error]: Unable to synchronize bootstrap for instance J00. The possible reasons: Kernel is older than expected.

Thanks.

[sunny_pahuja2]

Hi,

You can check SAP Note 1012486- Important Information on upgrading SAP Systems in HA-Setups.

Thanks

Sunny