Toby <tkb9@xxxxxxxxxxxx> writes: > With things connected as they are supposed to be, have you watched the > what the gw M5 (I assume it's a gw ) is doing? It may be taking the The original drawing did show it as a gateway. I inadvertantly left it out on the second one. I just realized too that both drawings are a little bit wrong in that M5 is also plugged to HUB1 or else none could reach it as gateway. What I labled M4 is really the 192.168.1.0 of M5. So in fact, one of the 4 on hub1 was the gateway M5. > packets & throwing them out on 192.168.0 net, which is the wrong place > of course. Trying the experiment Michael suggested.. removing M3 M4 (which is really one end of M5) removes that as a possibility. And in that configuration with only M1 M2 on HUB1. They cannot communicate. Puting M2 on a separate HUB2 and uplinking it to HUB1 works and all can communicate freely. > Personally, I would just remove completely the NETWORK & BROADCAST > lines out of the ifcfg scripts. You don't need them here - NETMASK > 255.255.255.0 perfectly describes the network on its own. That is a good point and good idea I think. I found at least three mistakes there on the 4 machines. However, it seems not to matter. Even with those errors (2 had mismatched broadcast addresses and 1 had a mismatched network address) M1 M3 M4/5 could communicate freely. Apparently the entries are not really used. -- Psyche-list mailing list Psyche-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/psyche-list