Issue with SAP system running on MSCS

[former_member209962]

Hi Friends,

From past two weeks we are facing one strange issue, users are complaining about system performance very slow, when we checked in the system if user is excuting any transaction first time system is taking very very long time and second time its coming faster , we understand that SAP say  ignore first run, but this is happening daily in between some hours for example today i run one t code first time its very slow and second run is fine  but after some time may be after 6 hour or next day again if i run the t code it will be slow and next time it will come fast.

this isssue was started with few users and spreading and suddenly all the work process came in running stage and user cannot log in to the system, hence we tried to restar the system and we notice that our micro soft cluster is having issue resoucres where not moving to Node2.

Hour network and windows admin find out that there is issue with virtual network between both the nodes please seee there comments

it was found that both your servers where trying to mount the cluster disks to themselves creating a grid lock.

In such an event both nodes assume that their ‎partner nodes are down and the primary node always locks the cluster resources and never allows second node to take the ownership of the resources.

This grid lock occurs when there is an issue with RPC services are not able to synch with net bios services or tcp/ip services or when you have tcp ip errors.

So I had to shut down both the servers un present and present he storage LUNs from both the ‎hosts allowing a re mapping of resources to the nodes.

After doing the above activities the node 2 was brought up with all your cluster resources with Oracle and sap services coming up

Kindly help us to fix this issue and let me know you need any logs

Thanks

Basis

[former_member209962]

Dear All,

Now MSCS is wokring fine and faliover is also ok

solution: We have restarted SAN switches.

Thanks

Basis

[hemanth2]

Hi Tabrayz,

Hope you are doing good.

This may sound lame, but did you run /nSGEN once? This will generate ABAP loads for programs, function modules/groups and other applications also like Business Server Page  applications (BSP applications).
The reason I am stating this is cause you mentioned that the first run is slow .

Thank you!

____________

Kind Regards,

Hemanth

SAP AGS

[former_member209962]

Hi Hemanth,

Thanks for you reply, yes SGEN is run when ever we do any patch upgarde

and system is not compiling just getting hang, this issue started from past 2 weeks and it happen randomly 

Thanks

Basis

[Sriram2009]

Hi

Is this any Windows OS patches & Security updates are missing? if possible you can raise the support ticket to Microsoft check the over all Windows OS & any San storage connectivity issue? because we are faced SAN connection issue.

Regards

Sriram

[Former Member]

it looks very much like your problem is caused on the Windows Server (OS - Version) and Storage Level.

Hence it would strongly recommend that you are contacting a more Windows Server / Storage oriented forum to get answers to your questions - or even better contact the hardware vendor who sold you this configuration.

As long there are problems with subordinate resources (storage, network) SAP cluster resources and 3rd party database resources can't work reliable.

kind regards

Peter

[former_member209962]

Hi Peter,

Thanks for you inputs  yes we are contacting vendor also to check the issue and mean while i am just trying to get as much as information so that we can solve this issue faster.

Thanks

Basis

[Former Member]

I had seen a similar problem once: we had a timeout during an isalivecheck of the SAP Cluster Resource. We could find out that sapstartsrv.exe was blocked during a very small write operation on a cluster disk belonging to the SAP cluster group.

The Failover Cluster Group was well defined and the SAP Service itself was depending on the Fileserver, which was depending on the disk. At the end of the day it points out, that the physical disk resource provided by Microsoft did only check the visibility of the physical disk but did never check whether there are I/Os possible to volumes located on it.

Therefore it is very important that the disk subsystem is running and functioning well.

If you have problems in IO on your clustered disks, your database may have bad responsse times of even hang, or other very hard to identify problems arise.

Every cluster resource developer is depending on the truth of operating system resources in the dependency stack.

regards

Peter