> > then ping another machine, then the gateway, before I can get a live > > network connection going. > > Can you try to diagnose this a bit more precisely?
I can try. This might be a drag, but I haven't used the Hurd in a few weeks because the new Mach package didn't detect my network card outright, and I haven't gotten around to reinstalling the old one or compiling a custom version for my system... I'll diagnose it soon. > Note that it is normal for a Hurd machine not to respond to pings right on > boot before anything on the machine has accessed the pfinet server (just > like a Linux/Unix machine before `ifconfig' has been run to set up the > network). The pfinet server, which is what listens to the Ethernet for IP > packets, is a translator set on /servers/socket/2, so it starts up (and > will should respond to pings) as soon as anything accesses that node. In a > multiuser boot this normally happens the first time when inetd starts up > and begins creating AF_INET sockets. Any other program run locally that > makes an AF_INET socket will start it up too (i.e. ping or anything else > that uses thet network). This is certainly not right after startup. > Can you give us a precise characterization of the situation when you see > this network nonresponsiveness? Is this just after boot? Singleuser or > multiuser? What else is running or has run? Does the period of > nonresponsiveness continue until you do something specifically > network-related (the pings you mentioned), or does the situation change > just given the passage of time? During a period of nonresponsiveness, what > does `ps -A' show? I have to perform more extensive testing to be of any use. This will take a couple of days, since I'm really busy lately. --Michael Bacarella
