oconnor@SCCGATE.SCC.COM (Michael J. O'Connor) (12/16/87)
Dave Mills reports that the two Umdnet gateways that we used for testing are X.25 machines. Allison Mankin reports that the two Mitre machines that we used are 1822. Since the Mitre machines exhibit the problem and the UMD machines do not, it is our early suspiciion that the problem only occurs when 1822 machines attempt connections to X.25 machines over asymetric routes. We are trying to discover the flavor of interface used by all the other machines we ran tests through. The fact that sending a single packet out the jammed VC will free it, might indicate that somewhere in the 1822 to X.25 conversion, a piece of code is patiently awaiting some indication that the circuit is alive before pumping the data on through. Dave suggested some more machines to test with (I'm sure at least one is a fuzzball) and we'll give them a try. Mike