Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!asuvax!ncar!elroy.jpl.nasa.gov!swrinde!zaphod.mps.ohio-state.edu!pacific.mps.ohio-state.edu!linac!unixhub!shelby!agate!ucbvax!pan.ssec.honeywell.com!thompson From: thompson@pan.ssec.honeywell.com (John Thompson) Newsgroups: comp.sys.apollo Subject: re: Help on tcp & crp... (please) Message-ID: <9102250140.AA03125@pan.ssec.honeywell.com> Date: 25 Feb 91 01:40:26 GMT Sender: daemon@ucbvax.BERKELEY.EDU Organization: The Internet Lines: 58 Ian (et.al.) -- > <> (text re-ordered for clarity) > Im having a few problems setting up the departments apollo network > We just got 6 DN4500's. Config is 6 x Dn4500 (16Meg 330Meg disks) > ethernet Domain/OS 10.2. The rest of the department is a mix of > suns,hp's,dec's and other misc unix boxes. > The main problem is the gateway onto the department > ethernet. One node is set up as a gateway, but when I run tcp/ip on the > apollo ethernet, the system seems to communicate in only one direction > ie I can rwho, and all the department node respond, but rlogin hangs > with no response. When I only run tcp on the gateway I can rlogin to > the remote net fine, however the rest of the apollo's dont see the remote > network... Any ideas? When I first read this, I said 'subnets -- no question.' When I got to the end and noticed the configuration, it changed to 'question.' You say that the Apollo's are gateway-ed onto the main department ethernet. Does this mean that they're on a separate piece of cable? If not, then there's no reason to have the gateway, and I suspect that your problem is being caused by having the nodes think they need to go through somebody who doesn't think they need to go through him. More likely, you have a second piece of wire, and your gateway node has 2 controllers, and is on 2 ethernets. If this is true, the likely problem is subnets. (In case you aren't familiar w/ it, subnets are simply a means of breaking up your one local tcp/ip network address into several sub-networks. The outside world treats them all as one network, but inside, you treat it as several. For instance, honeywell.com has a class B network address of 129.30.x.y. From the outside world, the "x.y" combinations give a potential of ~ 65000 systems on the same network. Since Honeywell is not all in one building (and because nobody wants 65000 nodes on one network!), it has been sub-netted. The 3rd byte (the 'x' one) is used to designate which sub network the host 'y' is on. Routing proceeds from the node to the gateway(s) to the recipient, and then back again. The most important thing to note in this is that _all_ hosts (Apollo and otherwise) on the 129.30 net must agree to use the same subnetting scheme. Otherwise, some recipients will think that the message came from their own network, and will not look for a gateway to send through. I suspect that either you don't have subnetting set up (in which case no packets are going across the gateway) or subnetting is set up only on the Apollo side (in which case packets go across, but the reply never gets routed back to your node). > Another minor thing is the crp program seems to hang after a few > commands! It then needs to be killed to clear the connection. No explanation for that one. I have had (almost) no problems w/ crp. The first thing to check out would be the patch tapes. Keep in mind that 'crp' uses the mailbox routines (mbx_helper) and the server_process_manager (spm) to get things set up. -- jt -- John Thompson Honeywell, SSEC Plymouth, MN 55441 thompson@pan.ssec.honeywell.com Me? Represent Honeywell? You've GOT to be kidding!!! Brought to you by Super Global Mega Corp .com