How to select Cluster Nodes in Communication channel

[rubanprasanth_s]

Hello Experts,

We are using PO 7.4 Single stack.
We have two different Server nodes ( Cluster Nodes ). Is there an option to choose a particular node ( every time ) for a particular communication channel..

For eg: Lets take a file sender communication channel named CC_SND_FILE

When ever a file is picked by this communication channel, randomely one of the cluster node is used . But i want to restrict this random pick and want to define which one to use.

Please let me know if this can be performed .

Thanks,
Ruban.S

[vadimklimov · Accepted solution]

Hi Ruban,

For file sender communication channel, you can restrict it to poll from the specific server node using advanced mode parameterization in the channel configuration - namely, parameter 'clusterSyncMode', where the parameter value shall refer to the cluster node on which polling job for this channel shall be scheduled. Details about this functionality and examples of its configuration can be found in SAP Note 2029410. Although this kind of configuration is technically possible, I see this as a very specific requirement in very limited carcumstances when you need to deviate from standard scheduler logic for polling channels and restrict polling to a specific node - hence, I support Patrick in questioning reason behind your requirement.

Regards,

Vadim

[weberpat · Accepted solution]

I can't view notes at the moment but if I'm not entirely mistaken, the clusterSyncMode parameter will allow you to either "lock" the channel to one cluster node or let the "scheduler" take over. I'm not aware of a setting that would let you control which particular node will be selected. If such setting exists, I would be very interested in learning about it 🙂

[former_member190293 · Accepted solution]

Hi Patrick!

As per note:

1. Add Additional Parameter with Name clusterSyncMode and Value node<clusterNodeID>[lock], where the <clusterNodeID> corresponds to the Cluster ID shown in SAP Management Console under AS Java Process Table (<instanceID>50 for server0, <instanceID>51 for server1, <instanceID>52 for server2, etc.) and lock is an optional suffix that defines how the system will behave in case the configured cluster node is not available or becomes unavailable, for example:

• clusterSyncMode=node9569350 will schedule the job to run at server0 of instance ID95693. In case the node is/becomes unavailable, at the next poll time the surviving nodes will report errors, the channel will go in error state (visible in Communication Channel Monitoring) and alerts can be triggered for the channel to automatically inform the PI Administrators about the problem. If the configured node does not come up and the channel is not reassigned to a different node in 10 minutes, new series of errors (and alerts) will be reported at the next poll time by surviving nodes. To receive the alerts the corresponding Alert Rules have to be configured for the Adapter Engine;
• clusterSyncMode=node9569352lock will schedule the job to run at server2 of instance ID95693. In case the node is/becomes unavailable, the rest of the server nodes will behave as in lockmode – the one that succeeds to acquire the channel lock will process the files.

  Be aware that after transporting/importing a channel configuration that uses the new clusterSyncMode from a different system, the value should be adapted to use clusterNodeID of existing cluster node in the current system.

Regards, Evgeniy.

[weberpat]

Just out of curiosity: what's your motivation for wanting to do that?

[mate_moricz]

Hi Ruben,

There is no possibility to choose the server node but some adapters' EOIO processing will happen only on node, for example JMS EOIO processing.

Best regards,
Mate

[Harish]

Hi Ruben,

AFAIK - you can not select any specific node to process a communication channel. This can be possible when you have decentral adapter engine where specific adapters are installed on specific nodes.

regards,

Harish