Friday, March 28, 2014

Recovering BPEL Activity Failures - WSO2 BPS


Read more about ODE activity failure and recovery from.

 

Method 1: Via WSO2 BPS Management Console 


In WSO2 BPS, BPEL failed activity recovery can be done using WSO2 BPS Management console.

To do this, it is required to enable ODE activityLifecycle events for a BPEL process. To do that modify process-event configuration in deploy.xml as follow.
(refer http://ode.apache.org/ode-execution-events.html for to see how you can enable ODE events for a BPEL process.)

Users can view activity failures for a BPEL process instance in the instance view page. 

1) Goto Instances -> click on of the instance ID to open the instance view for that instance.


 2) In the instance view, you can find failed activity/activities under activity informations. (See image). Also you can retry/cancel (ignore a failed activity) a failed activity using two buttons listed under Action column.





Method 2 - Via InstanceManagementService admin API



Also users can recover failed activities by using BPS InstanceManagementService. Unfortunately there is no UI functionality for this when ode events are disabled. (BPS 3.2.0 & older)


These are the steps, to retry activities using InstanceManagementService admin service.

1) Execute following SQL query on BPS database.

This will return the failed activities with corresponding BPEL process instance id. You will require following information to retry a failed activity.
  • INSTANCE_ID
  • ACTIVITY_ID

2) Set HideAdminServiceWSDLs configuration to false in /repository/conf/carbon.xml file.

eg:

3) Then start the BPS server.


4) Now create a SOAP-UI project using https://localhost:9443/services/InstanceManagementService?wsdl


5) Create a new request under recoverActivity operation. A sample request will be like this.


6) Then Authenticate soap-ui request by configuring authentication and security related settings.
For Basic Auth select Authorization type as preemptive, and give admin user name and password.

7) Then for each failed activity (results in step 1), send a recoverActivity request. Use "retry" action to retry activity, and use "cancel" action to cancel the activity retry and continue instance execution.