Sender RFC Channel seems to drop connection with SAP R/3

[Former Member]

Hi,

I have established an outbound SAP->XI RFC connection. It runs successfully for awhile but after an hour or two the RFC call from SAP starts failing because it can't find anyone listening on the specified destination. I have to go into the ID and reactivate the comm channel. to get things working aging What could be causing this? It has happened 5 or 6 times.

Regards,

Chris

[Former Member · Accepted solution]

Hi,

Try to monitor hardware resources and system load when it happens... We experienced this problem under high workload and no more memory available.

regards.

roberti

[Former Member · Accepted solution]

I have an update. The problem that is occurring is that it actually is using up the connections. But it only happens when the RFC is NOT being used for 1 hour. Everything works fine all day, then night comes and the RFC hangs saying that the connection is busy when it short dumps. I have to go into the comm channel and increase the max connections. This gets it started again, until the next hour of down time and so on. What could cause this?

[Former Member · Accepted solution]

Richard:

Even this happened to me when I was using File adapter.. Our adapter was stopping suddenly. when we dug deep into the problem we found out that it was doing when there no file for a long time to pick or when all the resources are at high peak. Then we have to use " Availability Scheduling" of the adapters to schedule instead of opening connections for a long time.

[Former Member · Accepted solution]

Hi,

try to go to RWB, component monitoring, adapter engine, communication channel monitoring. Search for your com. channel and look there what is its status. Also see processing details if there is some error log. Check admin information if there is some action registered - you would have there only one entry with "automatic initial start" registered there.

regards.

roberti

[Former Member · Accepted solution]

I found that the new server is in a locations whose firewall uses a TCPIP disconnect after one hour if there is no activity on the connections. To resolve this. The instance profile parameter gw\reg_keepalive was used at a 30 minute interval on the client systems to continously ping XI and keep the connection open!