rmallett@ccs.carleton.ca (Rick Mallett) (03/13/91)
I'm trying to get TCP/IP services to work on a network of 5 Apollo workstations running SR10.2 and I'm finding that the routing tables on the hosts disappear a few minutes after booting the gateway even though "netstat -r" on the gateway shows it has a full routing table. All machines use Apollo Token Ring for Domain services - the gateway has an ethernet card as well which is connected to the campus network. All 5 machines are setup to use name service (nmconfig -h hostent_bind) from Suns on the campus ethernet (i.e. named is not running on the ATR). There is a slight subtlety here. The machines are configured with an Apollo netid (set using netsvc) to isolate them from some other machines running SR9.7 on the same cable. I did this because I'm converting the net to SR9.7 but I don't want to convert the SR9.7 registry since it contains hundreds of obsolete accounts and I want to start over with a new registry. I've been moving the currently active users one by one onto the new system and it has all been working fine up to now. The point is that I don't want to install an Apollo Internet; in fact, I don't want my 5 machines to know anything at all about the other nodes. I do want my 5 nodes to see the real Internet. The manual "Managing Domain/OS and Domain Routing in an Internet" seems to assume that I want to set up an Apollo Domain network and I'm not sure how much of it is relevant to my problem. When I follow the instructions in "Configuring and Managing TCP/IP" I can't get the routing tables to stay around for more than a few minutes. If I boot the gateway node, it sends the Internet routing tables to the hosts which then function normally for about 10 minutes. After that "netstat -r" just hangs after printing the header or I get a message that no routing tables were found. If I sit on the gateway machine "netstat -r" always shows a complete routing table and I can ftp and telnet from there to anywhere in the world. Summary: 1. ATR network with netid set to "e1" running SR10.2 2. Gateway configured with RING and ETH802.3_AT interfaces. 3. routed invoked with "-f -q" on hosts and "-f" on gateway. 4. nmconfig -h hostent_bind on all nodes to use campus ethernet name service. Requirement: Advice on how to configure an Apollo Token Ring network to use TCP/IP services to access the Internet Internet (i.e not a Domain Internet). Any advice on how to proceed to debug the problem myself would also be greatly appreciated.
randall@bcstec.boeing.com (Michael Randall) (03/20/91)
In article <1991Mar13.144404.20653@ccs.carleton.ca>, rmallett@ccs.carleton.ca (Rick Mallett) writes: > I'm trying to get TCP/IP services to work on a network of 5 Apollo > workstations running SR10.2 and I'm finding that the routing tables > on the hosts disappear a few minutes after booting the gateway even > though "netstat -r" on the gateway shows it has a full routing table. > All machines use Apollo Token Ring for Domain services - the gateway On our hosts we don't use the routed. We have one command in /etc/rc.local "route add default ip.address.of.gateway.domain.ring 1". This tells the hosts that they should ask the gateway for any routing information. If your hosts file has: 192.192.10.10 host_eth0 150.150.150.1 host_dr0 then rc.local should have: if [ -f /etc/route ]; then /etc/route add default 150.150.150.1 1 fi I haven't experimented with changing the net ID. But this is worth a try. It has solved our TCP/IP problems (similar to yours). -- -------------------- Peace Love & E-Mail ---------------------------------- Michael W. Randall | Phone: (206) 965-9557 randall@bcstec.boeing.com | This space for rent... ...!uunet!bcstec!randall | ...Inquire within.