CloudStreams in clustered environments [SuccessFactors]

Hi all,

what are the recommendations in terms of configuring CloudStreams Connections in clustered environments? In our current setup an internal Integration Server is publishing requests from SAP to UM. We have two Intergration Servers that use active CloudStreams connections to SuccessFactors. They both have triggers configured that subscribe to the SAP documents. In this environment the requests get distributed evenly to both Integrations Servers. Whichever trigger picks up the SAP documents first, will send the requests through to SuccessFactors.

However it seems like there is some issue with two active CloudStreams connectors as we get the following or similar errors, supposedly due to inproper session management:

2018-08-01 12:17:22 CEST [CLS.0101.0002E] Error occurred while invoking CloudStreams Transport Sender. Cloud connector service: ‘wm.cloudstreams.sap.successfactors.odata2.lookup.interactions:describeMetadata’, Provider: ‘WmSAPSuccessFactorsProvider’, Details: An IO or timeout exception is encountered for invoke ‘GET’ method for connection alias ‘xxx.successfactors.adapters.connections:SuccessFactors_EC_OData’. Check the network connectivity/proxy setting. Details: Read timed out
com.softwareag.cloudstreams.conn.api.ConnectionFactoryException: An IO or timeout exception is encountered for invoke ‘GET’ method for connection alias ‘xxx.successfactors.adapters.connections:SuccessFactors_EC_OData’. Check the network connectivity/proxy setting. Details: Read timed out

As soon as we disabled the trigger on one of the IS nodes, the error stopped showing up. We have set up the default settings for the Connection Management Properties of the CloudStreams Connections (Connection Pooling set to enabled, Session Managment set to none).

Please advise on best practices for CloudStreams in clustered environments.

Thanks,
Ocin

Hi Ocin,

CloudStreams does not have inherent support for Cluster, so the recommendations for a clustered set up are pretty much the same as a standalone installation.The timeout errors that you observe are they consistent or one off? As long as the responses to the Successfactor API requests are received by the Integration Server which triggered the request, there should be no issue

Thanks,
Tanuka