sap portal 7.01 high availability problem

[iecisa_oss]

Hello,

we have a sap portal 7.01 SP06 on high availability cluster. We have SAP resources (SCS instance and web dispatcher) and oracle resources (oracle BBDD and listener). On node 1 runs java central instance and on node 2 runs java dialog instance and enqueue replication server.

When we move SAP and oracle resources from node 1 to node 2 or from node2 to node 1 manually without stop the machines, we check the status of the central and dialog instances and sometimes the dispatcher and server0 of both instances restart automatically but sometimes stops and we have to restart manually.

Thanks.

[Former Member]

Hi ,

If SCS is getting stopped ,java server and dispathccer will be stopped because of lack of connectivity to message server.

If SCS does not fget started before the no of restart attempts get exhausted ,the j2ee server/disptcher will not come up.

May be you can write a script to start j2ee as well

[former_member146451]

Hello,

We have four cluster scripts to automatically move sap (SCS and web dispatcher) and oracle resources:

startORAepx.sh: It starts listener and oracle BBDD.

stopORAepx.sh: It stops oracle BBDD and listener.

startSCSepx.sh: It starts SCS and web dispatcher.

stopSCSepx.sh: It stops web dispatcher, free memory and semaphores of web dispatcher user and stops SCS.

On initial state, we have listener, oracle BBDD, SCS, web dispatcher and java central instance run on node 1. On node 2 run enqueue replication server and java dialog instance.

When we move sap and oracle resources from node 1 to node 2, cluster software hacmp execute stopSCSepx.sh and stopORAepx.sh on node 1 and execute startORAepx.sh and startSCSepx.sh on node 2. The java central instance, enqueue replication server and java dialog instance are out of the automatic scripts and when it stops we need to start manually. But sometimes on this situation, this two java instances not lost the connection and continue works. It's disconcerting.

When we move the resources, we check java instances with jcmon and sometimes the dispatcher and the server0 are stopped and sometimes are starting and later running.

Is it probably that a not enough memory problem happens?

Thanks.

[Benny]

Whether your resources start and stop automatically is not in the hands of the SAP system, but a function of your HA software. The moving and activation of software is a function of the clustering software.

Please check there.

The only exception is in case there is something wrong with Enqueue replication, then the whole server is restarted.

Regards,

Benny