Possible Timeout in C4C (Soap) Web Service / Broken Pipe

[JaySchwendemann]

Hi all,

we have a web service C4C (SOAP) --> Cloud Integration on CF --> Cloud Connector --> 3rd party server providing PDF services (SOAP)

We experience a "broken pipe" error as described in https://launchpad.support.sap.com/#/notes/3117499. In the note as a possible resolution it is adviced to increase the sender timeout.

How, if even possible can we increase the sender timout for a C4C consumer (sender) web service

Many thanks and kind regards

Jens

[JaySchwendemann]

PS: The broken pipe error in CI occurs at about 5 minutes so I would think it is set to 300 seconds in C4C

[Conrad]

We're having the same issue since some days. We tried changing a setting in the communciation arrangement ("Consumer Max Wait Time(s)") but this had no impact.

[JaySchwendemann]

Hi Conrad,

many thanks for pointing that out. I did not now that field yet 🙂

For us it is set to 0, so we might fiddle with this (e.g also decreasing it to say 240 secs and see if we hit the "Broken pipe" error at roughly 4 mins then). I will update here with the results

For reference, it's this value we are talking about
cc-consumer-max-wait-time.png

[Conrad]

I've received an answer from a colleague of mine therefore it's no official SAP statement but he had this issues earlier and it seems the load balancer of C4C is explicitly configured to cancel connections after 5 minutes and this configuration can not be changed. Maybe we can get some official statement via incident since I haven't seen any documentation on that behaviour.

The solution for us is to package the requests in smaller batches to be sure we're under 5 minutes. If this is not possible in your usecase you could transform the requests into an asynchronous call (e.g. by queuing in CPI if you're using it) but of course this means a major change in your architecture if you need to continue working with the response.

Cheers

[JaySchwendemann]

We do have an incident open for this currently. Unofficial guts feeling of the current SAP asignee of the incident also leans towards load balancer / proxy in outbound of C4C and he also mentioned the 5 mins (albeit I kinda pushed him to the 5 mins topic to be honest)

Currently we need to provide some Cloud Connector logs, so let's see if there's a different reason for the issue.

If load balancer would be the reason I also only see changing architecture because we need a "one shot" request that results in a PDF which seems to take quite long to complete.

Many thanks so far, I keep you posted