mooremj@UV4.EGLIN.AF.MIL ("Martin J. Moore") (02/16/90)
I'm new to this group, so please excuse me if these problems have been discussed before. Our facility is installing 3 ProNET-80 rings to connect 6 VAX 8600's (running VMS) and 6 PDP-11/84's (running custom executives of our own creation). We are using locally written drivers on both CPU types. Each ring connects to all 12 nodes (the reason we are using 3 rings is mainly for redundancy.) We are seeing two strange problems, and hope someone can give us some insight. Problem #1: Certain network configurations seem to cause us problems. When only two nodes are in the ring, it works fine. When three nodes are in the ring (*any* set of three nodes) we experience a large number of output format, output timeout, and ring not OK errors. Four nodes works pretty well, but not as well as two. Five nodes is pretty bad, but not as bad as three, and so on. At present we have only 7 nodes available (1 PDP and all the VAXen), and all 7 works fine. All three rings exhibit this behavior. Problem #2: Only one of our rings has this problem. The receivers (all of them) on this ring often, but not always, report odd byte count errors. It doesn't seem to matter which nodes, or how many nodes, are in the ring. There does seem to be a correlation with packet length: the longer the packet, the greater the frequency of ODB errors. Any suggestions or insights will be welcome. Thanks. Marty Moore mooremj@uv4.eglin.af.mil