Some users are reporting an error being returned from Entire-X : “WSAECONNRESET – Connection reset by peer. This occurs when an established connection is shut down for some reason by the remote computer.”

Any ideas what this actually means and how to prevent it? Any reason Entire-X doesn’t just handle it itself and retry the call?

Client is Windows. Server is z/os. We recently upgraded server to 7.3.2.06 but clients are still using 7.2.1.11.

The users claim this has only been a problem since this upgrade but they have no firm proof and I am sceptical. They think it happens after some length of inactivity but cannot tell me how long that needs to be to cause the error.

WSAECONNRESET is a rather low-level error message from the TCP/IP stack. Nothing EntireX specific.

If it happens after some length of inactivity: talk to your network people if they have a router/bridge/proxy/firewall/… or some other network hardware which closes IP connections after a certain timeout. Also in case you have Broker 7.3: check your attribute if you have defined CONNECTION-NONACT with a value greater 0.

The mainframe broker has “CONNECTION-NONACT = 1H”. Is this a new parameter with 7.3? What would the equivalent value be to get the same state as was the case under 7.2? Is there any way to overrride this on the windows client end as a test since we do not have control of the mainframe broker end due to outsourcing?

Yes, this is a new parameter in 7.3: http://techcommunity.softwareag.com/ecosystem/documentation/crossvision/eli/relnotes/newVersion.htm#relnotes_transportEnhancements

If you do not want to close inactive sockets after 1h remove the parameter or set it to 0: http://techcommunity.softwareag.com/ecosystem/documentation/crossvision/eli/adminGeneral/attributeFile.htm#CONNECTION-NONACT_tcp_ip

You can only change this value in the broker attribute file.