Hi everybody,
i’ve build a SAP BODS Real-Time Service in Version 12.2.2.0.
It works fine, but on error the service crash and the Access Server need a long time to start a new instance of the service.
Following Scenario:
I let the service fail with a database error because of a unique key constraint violation. After that all dataflows and workflows get completed successfully, but this need a long time. In this time (max 3 minutes) the Access Server doesn’t receive/progress any other request from the client for this service. After the end of the processing time the client get a Process Timeout Error. After these 3 minutes the Access Server start e new instance. The problem is that the Access Server doesn’t queue the request until the new instance get started.
As workaround I tried to start min 5 Instances of the service. But with 5 errors in 3 minutes I had the same problem of course. Sometimes it seems that the Access Server trys send a request to the failed service before all other 4 Instances recieve
the next request. So the error occurs earlier.
When I stop a service in the Management Console the service need only 1 second to shut down.
This Scenario is reproducible.
Please find attached the error and trace log of one request example, also the log of restart the service over the Management Console.
Why isn’t the restart so fast on failure as restart over the Management Console or how can I configure the service, that all request get queued?
In this case we lost important data.
This behavior only occurs when the job fail with a DB error. When the job fail with an BODS raise_exception the service gets immediately restart. Maybe the Access Server lost the Connection to the service?