I have a scenario where two asynchronous messages are sent from system A via PI to system B (both ABAP systems) and each triggers a confirmation message that goes from B via PI to A - or should go, one of them gets stuck with a red flag in system B and doesn't make it to the PI.
The error message is Web service processing error; more details in the web service error log on provider side
The calls from A to PI and from PI to B use XI. The successful call from B to PI (and from there to A) uses XI. The one that gets stuck (from B to PI) uses WS - and I cannot figure out why.
I checked the configuration in PI and sxi_cache in B. I assumed one or the other or both would show a sender agreement with the WS channel (for the failing confirmation message from B to PI) and something would be misconfigured in it. But there is none! No sender agreement in PI and none in sxi_cache, and the XI cache is up to date.
So why does one of the calls from B to PI suddenly use the WS channel, even though there is no sender agreement? Which other configuration could cause this?
Also, what is the web service error log / how do I check it? It must be on PI, as this is the provider side in the call from B to PI.