dean@bragi.cs.cornell.edu (Dean Krafft) (05/16/89)
I have a feeling that this problem was addressed a number of months ago on this newsgroup, but naturally I didn't need to worry about it at the time and can't find anything about it now. A secondary question then: Does anyone keep archives of this newsgroup? If so, where? The main problem: I have CAP 4.0 software providing mostly printer service for a bunch of Unix machines. This goes through a KFPS-2 running KIP-1/88. We have gotten a bunch of Mac IIs with Ethertalk cards, and I'd like to have them be able to do Appletalk through the Kinetics to the Laserwriters. According to Kinetics, K-Star is supposed to provide all the functionality of KIP, and in addition support direct Appletalk bridging onto the Ethernet. To run K-Star, I have gotten the Kinetics 2 -> 2/U upgrade. Unfortunately, when I try to switch to the new box, the KIP gatewaying function vanishes. I can access the new box running K-Star (and the same atalkatab configuration file) from the Macs or from the UNIX side, but when I run the CAP look program on my unix host, I only see the Kinetics box, and none of the Laserwriters. The only thing different about the listing for the Kinetics box is that it reports itself as "IPADDRESS" rather than "IPGATEWAY" the way the KIP-code box did. So, does anyone know what's going wrong? I've called Kinetics, and gone through the configuration with them, and they can't see anything wrong. Any help would be much appreciated. Dean Krafft Director of Computing Facilities Computer Science Dept. Cornell University (607) 255-9214 dean@cs.cornell.edu
kenton@athena.mit.edu (Kenton C. Phillips) (05/18/89)
In article <27890@cornell.UUCP> dean@cs.cornell.edu (Dean Krafft) writes: >.... >To run K-Star, I have gotten the Kinetics 2 -> 2/U upgrade. > >Unfortunately, when I try to switch to the new box, the KIP gatewaying >function vanishes. I can access the new box running K-Star (and the same >atalkatab configuration file) from the Macs or from the UNIX side, but >when I run the CAP look program on my unix host, I only see the Kinetics >box, and none of the Laserwriters. The only thing different about the >listing for the Kinetics box is that it reports itself as "IPADDRESS" >rather than "IPGATEWAY" the way the KIP-code box did. > > Dean Krafft > Director of Computing Facilities > Computer Science Dept. > Cornell University > (607) 255-9214 > dean@cs.cornell.edu I have a very similar problem. From the net with the 2/U box I can see machines on other AppleTalk nets (which happen to be on 2 boxes running KIP), I can print to LaserWriters on those other nets. I can even get CAP Appleshare access to the unix machines on the ethernet. But NCSA Telnet cannot get a dynamic IP address from the 2/U. If I specify a static one, I cannot telnet to the unix machines - I get an error message "host or gateway not responding". Like Dean, my 2/U box identifies itself as "IPADDRESS" rather than "IPGATEWAY". Also like Dean's case, Kinetics was less than helpful. They seemed to feel that, if I bought their replacement for telnet (at $100 a pop) then everything would work fine. I am skeptical, as it's clearly the gateway which does not know it's a gateway, and not the telnet program. It might be noted that atalkad's log file does record that the 2/U has requested information, just the same as for the KIP boxes. And, as I said, Macs on the 2/U's net can get to Appleshare files on the unix machines. (As I understand it, this doesn't require IP, though.) Does anyone have any suggestions about configuring this thing? [As an aside, and to vent frustration, I didn't order 2/U's, just 2's. I spent months trying to get these things to work berfore I even knew there *was* such a thing as a 2/U. I tried putting old 3.0 ROMs in the 2/U, only to find out that they're incompatible. Otherwise I'd just be using the KIP software in all my gateways, since I know how to make that work.] -- Kenton C. Phillips kenton@space.mit.edu
troym@amtfocus.UUCP (Troy Monaghen ) (05/22/89)
I had what I think is a similar problem when I first set up my FastPath. My unix machine does not currently support kip so I had to start out with K-STAR. I spent several days playing with various combinations of Appletalk Zone and address names in the K-STAR configuration, but all I could see with atlook was the IPADDRESS reply from the FastPath. It turned out that every configuration I tried had the LocalTalk and Ethernet AppleTalk Net Number the same. As soon as I made them different... voila! All of a sudden all of my AppleTalk entities appeared and everything worked from then on! So my current working configuration has 3 different Appletalk net numbers (one for LocalTalk, one for Ethernet, and one for UDP). Since I am relatively new to the appletalk world, I don't really know what constraints there are to AppleTalk Net's and Zones, and in the three or four time I went searching through the Kinetics Manual I did not see any guidelines. Does anybody know of any guidelines for AppleTalk net numbers and FastPath's? -- Troy Monaghen, Motorola, Inc. chg.mcd.mot.com!amtfocus!troym General Systems Group - AMT uunet!harper!amtfocus!troym
farrell@EREBUS.STANFORD.EDU (Phil Farrell) (05/23/89)
> I have a very similar problem. From the net with the 2/U box I can see > machines on other AppleTalk nets (which happen to be on 2 boxes running > KIP), I can print to LaserWriters on those other nets. I can even get > CAP Appleshare access to the unix machines on the ethernet. But NCSA > Telnet cannot get a dynamic IP address from the 2/U. If I specify a > static one, I cannot telnet to the unix machines - I get an error > message "host or gateway not responding". > Kenton C. Phillips > kenton@space.mit.edu I have experienced the same problem as Phillips. MacIP cannot get an IP address assignment from my upgraded KFPS-1 running K-STAR (version 5.03, PROM version 4.2), although AppleTalk protocol traffic goes through the box fine (Chooser works to locate LaserWriters on the same or other zones, AUFS volumes can be mounted, etc). I did discover a work-around, which I will mention after first explaining my network situation. I have seven separate LocalTalk (or Phone-Net) networks connected to our campus ethernet via Kinetics FastPaths. I had one KFPS-1, three KFPS-2s, and one KFPS-3, all running KIP (9/87 version); plus two KFPS-4s (both with the latest PROM 4.1) running KSTAR (latest version 5.03). All but one of the FastPaths connected to the same ethernet cable; the seventh (a KFPS-2) connects to a cable in another building accessible by a standard IP router. I assigned the same zone "GEO" to six of the FastPaths (the LocalTalk segments), ethertalk on both ethernet cables, and IPtalk (appletalk-in-IP) on both ethernet cables. One KFPS-4 is in a separate LocalTalk zone "SRB". I run the atalkad daemon (6/88 version) on my VAX to administer all seven boxes from one /etc/atalkatab. This all worked dandy - everthing could see everything else, MacIP worked on all LocalTalk segments, etc. When my venerable KFPS-1 crapped out and had to be shipped back for repairs in April, I decided to go ahead and have it upgraded to a KFPS-1U and run K-STAR on it. I plugged it in, loaded K-STAR, and used the same /etc/atalkatab entries as before. As I said above, AppleTalk traffic worked fine, but now MacIP would not work on the LocalTalk segment served by that FastPath, claiming that it could not get an IP address assignment. Our Networking group here at Stanford put a network "sniffer" on the ethernet cable and watched packets: when MacIP tried to get its IP assignment, the KFPS-1U would send out NBP lookups for gateway service to all the other FastPaths and IPtalk nets it knew about, it would get appropriate replies, but it would fail to finally tell the Mac what to do. Here was my workaround: I put the KFPS-1U into a separate LocalTalk zone, all by itself. Now MacIP works fine. This is a bit of an inconvenience for our naive users on that LocalTalk segment, who now have to select a different zone in Chooser to see the LaserWriters, etc., and it would complicate my configurations if I wanted to spool print jobs from my Vax to any LaserWriter served by the KFPS-1U (none yet), so I would like to see Kinetics fix this problem and let my KFPS-1U coexist in the same zone with my other boxes. However, as Phillips mentioned, Kinetics has not been very helpful (telephone support says "Gee, we've never seen that" with the implication that therefore, it must not exist). Personally, I think there is a bug in the PROM 4.2 that is used in the upgraded KFPS-1s, 2s, and 3s, because this problem does not affect my KFPS-4s with PROM 4.1.
jmg@cernvax.UUCP (john gerard) (05/30/89)
I have a similar sort of problem, but with K-Star replacing the traditional Kinetics IP subnetting approach. I run the simplest possible K-Star configuration: LoaclTalk side Net 36 on a distinct zone, Ethernet side Net 1 called Ethernet (original, eh?), FP IP address 128.141.30.36, no Administrator, name server on 128.141.200.5, 32 static addresses, no options. I can see other zones happily in the chooser, and they can see me. However, when I try to use either NCSA telnet or TCPort telnet (with IP address 128.141.36.1) I never see any packet go out on the Ethernet (verified on the Lanalyzer). Downloading the FP gives in the diagnostics box the message that IP routing is off: funny, but I don't see any option to turn it on. Having read the documentation three times, I am still unsure how the FP ought to decide what IP addresses it is meant to serve on its own LocalTalk. I have that worrying feeling about missing something obvious, but am now at the stage that I don't care if people think I am stupid. ps. am also trying to make a gatorbox work, with about as much success. This is obviously not my month! pps. we don't use subnetting, so IP subnet Mask is put as 0.0.0.0, as is broadcast address Mike Gerard -- _ _ o | __ | jmg@cernvax.uucp | | | | _ / \ _ __ _ __ _| jmg@cernvax.bitnet | | | | |_) /_) | __/_) | (___\ | (_/ | J. M. Gerard, Div. DD, CERN, | | |_|_| \_/\___ \__/ \___| (_|_| \_|_ 1211 Geneva 23, Switzerland
wnn@DSUNX1.DSRD.ORNL.GOV (W. N. Naegeli) (05/31/89)
Mike Gerard writes: >>>>>>>> I run the simplest possible K-Star configuration: LoaclTalk side Net 36 on a distinct zone, Ethernet side Net 1 called Ethernet (original, eh?), FP IP address 128.141.30.36, no Administrator, name server on 128.141.200.5, 32 static addresses, no options. I can see other zones happily in the chooser, and they can see me. However, when I try to use either NCSA telnet or TCPort telnet (with IP address 128.141.36.1) I never see any packet go out on the Ethernet (verified on the Lanalyzer). <<<<<<<< We had the same problem. In IP subnetting the FastPath has two separate IP addresses for the Ethernet side and the LocalTalk side. In K*Star it has only a single address. The IP addresses that you use on the LocalTalk side must be within the same contiguous range as the FastPath's address. I.e. if you specify 32 static addresses, you do not only specify the number of addresses (maximum) that may be used on the LocalTalk side, but you are also reserving the addresses themselves. In your case they are between 128.141.30.37 and 128.141.30.68. If you use anything else, you can communicate between nodes on the local LocalTalk, but not beyond the Kinetics box. Incidentally, if you specify say 16 static and 16 dynamic addresses, you have available addresses 128.141.30.37 through 52 for statically configured telnet invocations, etc. Dynamic assignments work backwards within the range, the first request gets 128.141.30.68, the next ...67, etc. down to ...53. Needless to say, that we had to revamp our IP address allocation scheme before we could switch from IP subnetting to K*Star. As usual, the Kinetics manuals were of little help, and only during the third call to Kinetics Technical Support did my tenacious prodding lead to a mutual discovery process in which their technician probably learned as much as I. Now, this is my penny's worth of wisdom. Perhaps there is another solution, but at least the people I talked with at Kinetics did not know, and this it the only thing I found that works. Wolfgang N. Naegeli Oak Ridge National Laboratory wnn@dsunx1.dsrd.ornl.gov (128.219.96.46) (615) 574-6143