jhh1@ra.MsState.Edu (Jim Harfst) (11/21/90)
Help! I am trying to telnet from a Mac IIcx in one zone through a Fastpath 4
in another zone and I can't find any zones on the Ethernet backbone. Here
is my setup:
MAc IICX ------------Shiva Netbridge------------Fastpath ===============Host
ATAlk Atalk Ethernet
The Mac IIcx can't seem to find the host. I am using Stanfords MacIP 4.0
with MacTCP. I have the Fastpath set up for static addressing and everything
is Phase II. Any machine *between* the Netbridge and the Fastpath has no
problem telneting or FTPing, etc.
I am beginning to get desperate. Help!
Jim Harfst
Mississippi State Universitymst@mx.csun.edu (Michael Temkin) (11/21/90)
In article <jhh1.659135509@ra> jhh1@ra.MsState.Edu (Jim Harfst) writes: >Help! I am trying to telnet from a Mac IIcx in one zone through a Fastpath 4 >in another zone and I can't find any zones on the Ethernet backbone. Here >is my setup: > > > > >MAc IICX ------------Shiva Netbridge------------Fastpath ===============Host > ATAlk Atalk Ethernet > > >The Mac IIcx can't seem to find the host. I am using Stanfords MacIP 4.0 >with MacTCP. I have the Fastpath set up for static addressing and everything >is Phase II. Any machine *between* the Netbridge and the Fastpath has no >problem telneting or FTPing, etc. > >I am beginning to get desperate. Help! > > > >Jim Harfst >Mississippi State University FastPaths and GatorBoxes ONLY BRIDGE IP TRAFFIC THAT ARE WITHIN THEIR STATIC OR DYNAMIC RANGES. If you IICX's IP address is not maintained by the box, IT WILL NOT forward it. (Sorry for all the caps, but I am trying to stress a point in stupidity on the part of Kinetics and Cayman.) Mike. -- Mike Temkin mst@csun.edu Cal. State U. Northridge, School of Engineering and Computer Science Voice phone: (818) 885-3919
powsner@csb1.nlm.nih.gov (Seth M Powsner) (11/21/90)
I think there is another problem-- at least under K Star 7.0 Static / Manual addressing results in some sort of acknowledgement request from the FastPath to AppleTalk Node 255. This does not get routed by the NetBridge back to the Mac's home zone. This may stem from the original KIP spec. Might be considered a feature-- if you're counting on manually assigned IP Addresses belonging to a specific machine, you could be assured the address wasn't being usurped by someone outside the FastPath's own zone. On the other hand.... I've had no trouble with the configuration you describe (Mac----NetBridge---FastPath---local ethernet - - - beyond) so long as Dynamic / Server IP Address assignment was used AND Privacy was OFF on the NetBridge. Seth Powsner powsner@nlm.nih.gov powsner@yalemed.bitnet
jhh1@ra.MsState.Edu (Jim Harfst) (11/21/90)
In <1990Nov21.034809.23021@csun.edu> mst@mx.csun.edu (Michael Temkin) writes: [stuff deleted] >FastPaths and GatorBoxes ONLY BRIDGE IP TRAFFIC THAT ARE WITHIN THEIR >STATIC OR DYNAMIC RANGES. If you IICX's IP address is not maintained >by the box, IT WILL NOT forward it. (Sorry for all the caps, but I am >trying to stress a point in stupidity on the part of Kinetics and Cayman.) >Mike. >-- >Mike Temkin >mst@csun.edu >Cal. State U. Northridge, School of Engineering and Computer Science >Voice phone: (818) 885-3919 Our Fastpath is set up for 20 static addresses. The IIcx's address is one within this range. Also, the Gateway address in MacTCP is set up for the Fastpath. All of the Macs on our net are set up the same way (none outside the IP address range). The only difference is that the IIcx is in a different zone than the Fastpath. Also another Mac in a third zone off the first zone has the same problem. Jim Harfst Mississippi State University
corwin@Apple.COM (Paul Frommeyer) (11/22/90)
jhh1@ra.MsState.Edu (Jim Harfst) writes: >Help! I am trying to telnet from a Mac IIcx in one zone through a Fastpath 4 >in another zone and I can't find any zones on the Ethernet backbone. Here >The Mac IIcx can't seem to find the host. I am using Stanfords MacIP 4.0 >with MacTCP. I have the Fastpath set up for static addressing and everything >is Phase II. Any machine *between* the Netbridge and the Fastpath has no >problem telneting or FTPing, etc. >I am beginning to get desperate. Help! Assuming you're using KSTAR 8.0, here's the deal: As you may know, there are two kinds of AppleTalk DDP packets, short and long. Long packets are designed to be used on an extended network (one with routers), and contain lots of good stuff about the network numbers of the source and destination networks. Short DDP packets do not contain the network number fields. Now. The FastPath and MacTCP (for reasons unknown to me) like to use short DDP packets for NBP BrRq (name lookup). So, any router adhering to the letter of Inside AppleTalk specs and not forwarding short DDP packets, will cause the IPGATEWAY lookup from MacTCP to appear to "blackhole" with no response. This gets amusing when using dynamic KIP, because the Mac will get its address (MacTCP uses long DDP packets for that), but then the gateway lookup breaks... I think this all came about because originally no one seems to have ever intended KIP to work across networks and zones. But I digress. We use Apple Internet Router software (suprise, suprise), and luckily for our FastPath users this will forward short DDP packets. Just how the router software is able to sort this out isn't entirely clear to me (I don't write any of this stuff, I just have to use it). In any event, it sounds to me like the NetBridge, though a fine product, isn't passing on short DDP packets. All of this assumes that your MacTCP setup is in order. You must have encapsulation (IP-IN-DDP KIP) selected (but then you're not on Ethernet anyway), and you must have correctly configured your network numbers, network mask, and subnet value (very important) so that they _agree across the board_ with the FastPath's idea of what these should be. As another poster pointed out, the FastPath will only provide encapsulation service for hosts within its address range. If any one of these is out of sync, you'll be S.O.L. You must have the zone in MacTCP set to that *of the FastPath*, NOT the Macintosh. If you are subnetting your LocalTalk (I never understood the need for that...), you are in for some additional headaches in configuring the FastPath. Your best bet is to do some network analysis and see just what dialogue, if any, is occuring between the FastPath and your Macintosh. I've found the Watch network analysis application (available from the Info-Mac archives at sumex-aim.stanford.edu) to be pretty good at debugging KIP service problems, though it's riddled with bugs and idiosyncrasies. This is really the best way to go; no two networks are alike, and no amount of advice is as good as a thorough site-specific analysis given the unique nature of network problems. Hope this was of some help to you and other net.readers wrestling with KIP service. It'd be nice if we heard from Shiva on this, since they make the NetBridge and (now) the FastPath... Cheers, Paul >Jim Harfst >Mississippi State University -- Paul "Corwin" Frommeyer "The whole damn world's gone to hell since Network Sorcerer and Telecomm Hacker David Lee Roth left AT&T" Apple Computer Incorporated Internet: corwin@apple.com UUCP: apple!corwin printf("Disclaimer:%s\n",common&&sense?"My opinion, not Apple's":"No comment");
LMOKUN@cc.utah.edu (11/24/90)
In article <jhh1.659135509@ra>, jhh1@ra.MsState.Edu (Jim Harfst) writes: > Help! I am trying to telnet from a Mac IIcx in one zone through a Fastpath 4 > in another zone and I can't find any zones on the Ethernet backbone. Here > is my setup: > > MAc IICX ------------Shiva Netbridge------------Fastpath ===============Host > ATAlk Atalk Ethernet > > The Mac IIcx can't seem to find the host. I am using Stanfords MacIP 4.0 > with MacTCP. I have the Fastpath set up for static addressing and everything > is Phase II. Any machine *between* the Netbridge and the Fastpath has no > problem telneting or FTPing, etc. > > I am beginning to get desperate. Help! > > Jim Harfst > Mississippi State University One thing to check: Is the Netbridge an "AppleTalk Phase1 router"? If so, then is the FastPath configured as a "transition bridge" (routing AT1 as well as AT2)? The Kinetics FastPath manual says that if there is a Phase 1 router or bridge on a LocalTalk [sic] net, then any FastPath connected to that net *must* be configured as a transition bridge, not as just an AT2 router. Larry Okun Dept. of Biology University of Utah larry@biology.utah.edu lmokun@cc.utah.edu
brad@CAYMAN.COM (11/24/90)
>> Date: 21 Nov 90 14:09:46 GMT >> From: usc!zaphod.mps.ohio-state.edu!uakari.primate.wisc.edu!ra!jhh1@apple.com (Jim Harfst) >> Subject: Re: Need help with Telneting across Zones with Fastpath >> >FastPaths and GatorBoxes ONLY BRIDGE IP TRAFFIC THAT ARE WITHIN THEIR >> >STATIC OR DYNAMIC RANGES. If you IICX's IP address is not maintained >> >by the box, IT WILL NOT forward it. (Sorry for all the caps, but I am >> >trying to stress a point in stupidity on the part of Kinetics and Cayman.) >> >> >Mike Temkin >> >mst@csun.edu >> >Cal. State U. Northridge, School of Engineering and Computer Science I will go out on a limb here and make some assumptions. 1) is that the Mac IIcx is in a different zone than the fastpath or gatorbox. 2) is that you won't call me stupid, as I wrote the code that does this in the gatorbox and am working with the IETF workgroup to standardize the protocol. The problem may be that given the current "psuedo standard" protocol for encapsulating IP inside AppleTalk, statically assigned addresses are not accessible from zones other than the localtalk zone of the gateway. Why? The gateway needs to establish a mapping between the appletalk address of the macintosh and the macintosh's IP address (just like ARP does with hardware addresses on ethernet). With static addressing, no conversation occurs between the macintosh and the gateway. Note: dynamically assigned addresses will work in this case. This is because the macintosh will "talk" to the gateway and "ask" for an IP address. At the time of "asking", the gateway can remember the appletalk source address and complete the ip address to appletalk address mapping. Statically assigned addresses work on the local zone becase the gateway will "arp" using NBP in the local zone for the mac's IP address. Mac's not in the local zone (like your mac, behind a netbridge) will not see these lookups and will not respond. Two simple enhancements can solve this (warning - oversimplification of a solution comming) 1) is to have the gatorboxes and fastpaths configured with a "list of zones" to search for statically assigned IP addresses. You could then add the zone name which you Mac IIcx resides to this list and be happy. 2) is to extend the protocol to include a way for statically assigned mac's to "talk" to the gateway to establish their mapping. I would not be suprised if both solutions arrive at around the same time. (and of course, everything I have said here may be wrong - feel free to send me e-mail if more explaination would help) -brad
resnick@cogsci.uiuc.edu (Pete Resnick) (11/25/90)
mst@mx.csun.edu (Michael Temkin) writes: >In article <jhh1.659135509@ra> jhh1@ra.MsState.Edu (Jim Harfst) writes: >> >>MAc IICX ------------Shiva Netbridge------------Fastpath ===============Host >> ATAlk Atalk Ethernet >> >FastPaths and GatorBoxes ONLY BRIDGE IP TRAFFIC THAT ARE WITHIN THEIR >STATIC OR DYNAMIC RANGES. If you IICX's IP address is not maintained >by the box, IT WILL NOT forward it. This is not Jim's Problem. It is not IP that is at issue here at all, but Appletalk. And I assume that he has the static and dynamic ranges set properly on his Fastpath (You did check that, right Jim?). Either way, your answer says nothing to his problem. > (Sorry for all the caps, but I am >trying to stress a point in stupidity on the part of Kinetics and Cayman.) So now that you found somewhere to grind your axe, what is the problem here. I certainly don't see why it is STUPID of bridges to do this!! Aside from being to spec, it might be quite nice to control which IP addresses your Macintosh's get and certainly is unacceptable to forward all IP traffic to the localtalk side. If the problem you are complaining about is that you need to specify a range and cannot use single addresses (i.e. configured from the Mac or the like), maybe you want to mention this to one of the companies in question. Cayman has been very helpful and the person I have been speaking to has been talking to the engineers about this as a down the road design change. Stupidity is a stupid charge. pr -- Pete Resnick (...so what is a mojo, and why would one be rising?) Graduate assistant - Philosophy Department, Gregory Hall, UIUC System manager - Cognitive Science Group, Beckman Institute, UIUC Internet/ARPAnet/EDUnet : resnick@cogsci.uiuc.edu BITNET (if no other way) : FREE0285@UIUCVMD
mst@mx.csun.edu (Michael Temkin) (11/26/90)
First, I would like to apologize for the length of this article, but in order to not be taken out of context, I have supplied the original posting. Mike. In article <1990Nov25.000830.9840@ux1.cso.uiuc.edu> resnick@cogsci.uiuc.edu (Pete Resnick) writes: >mst@mx.csun.edu (Michael Temkin) writes: >>FastPaths and GatorBoxes ONLY BRIDGE IP TRAFFIC THAT ARE WITHIN THEIR >>STATIC OR DYNAMIC RANGES. If you IICX's IP address is not maintained >>by the box, IT WILL NOT forward it. > >This is not Jim's Problem. It is not IP that is at issue here at all, >but Appletalk. And I assume that he has the static and dynamic ranges >set properly on his Fastpath (You did check that, right Jim?). Either >way, your answer says nothing to his problem. > >> (Sorry for all the caps, but I am >>trying to stress a point in stupidity on the part of Kinetics and Cayman.) > >So now that you found somewhere to grind your axe, what is the problem >here. I certainly don't see why it is STUPID of bridges to do this!! >Aside from being to spec, it might be quite nice to control which IP >addresses your Macintosh's get I agree whole-heartedly in regards to dynamically assigned ones... > and certainly is unacceptable to forward >all IP traffic to the localtalk side. If the problem you are complaining >about is that you need to specify a range and cannot use single addresses >(i.e. configured from the Mac or the like), maybe you want to mention >this to one of the companies in question. Cayman has been very helpful >and the person I have been speaking to has been talking to the engineers >about this as a down the road design change. Stupidity is a stupid charge. > >Pete Resnick (...so what is a mojo, and why would one be rising?) You are partially correct. Here is the original posting... >Help! I am trying to telnet from a Mac IIcx in one zone through a Fastpath 4 >in another zone and I can't find any zones on the Ethernet backbone. Here >is my setup: > >MAc IICX ------------Shiva Netbridge------------Fastpath ===============Host > ATAlk Atalk Ethernet > >The Mac IIcx can't seem to find the host. I am using Stanfords MacIP 4.0 >with MacTCP. I have the Fastpath set up for static addressing and everything >is Phase II. Any machine *between* the Netbridge and the Fastpath has no >problem telneting or FTPing, etc. > >Jim Harfst 1. If you read Jim's posting, he is trying to use MacIP to telnet and such, as far as I know this has to do with IP. True, his problem might be the NetBridge, phase II, or a dozen other factors, but to say it has nothing to do with IP is clearly incorrect. 2. If you read what I wrote, it says that if you give an IP address (static) to a Mac that is on LocalTalk behind a box, you had better make sure that the address is within the range managed by the box, otherwise the box will ignore any traffic from that host. I never implied nor did I mean to imply that the box should let all traffic pass onto the LocalTalk segment from the Ethernet segment. But IP traffic that is not local to the LocalTalk segment should be sent on through to the Ethernet side. 3. I never stated that the people who created the gateways were stupid, I myself have great respect for them as I could not do their job (and besides, without them we would be in even worse shape). What I claimed the problem was, was only taking care of traffic within its own range of addresses. I fully understand the fact that the boxes of that caliber (no slight intended) would be swamped if 253 nodes tried to talk at the same time through it. Even the recommended limit of 60 nodes can swamp them. But limiting them so that they ignore addresses outside of their range is (in MY opinion) a poor design choice. As Pete pointed out in his response, the service Cayman gives is excellent! I have called them with problems and they are very helpful and exceptionally quick to return my calls. Can't say anything about Shiva as I have not had to call them (although Kinetics was good, not as good as Cayman, but at they did try :-). Besides, if I did not like the GatorBox, I wouldn't be ordering two more... Mike. > (Sorry for all the caps, but I am >trying to stress a point in stupidity on the part of Kinetics and Cayman.) So now that you found somewhere to grind your axe, what is the problem here. I certainly don't see why it is STUPID of bridges to do this!! Aside from being to spec, it might be quite nice to control which IP addresses your Macintosh's get and certainly is unacceptable to forward all IP traffic to the localtalk side. If the problem you are complaining about is that you need to specify a range and cannot use single addresses (i.e. configured from the Mac or the like), maybe you want to mention this to one of the companies in question. Cayman has been very helpful and the person I have been speaking to has been talking to the engineers about this as a down the road design change. Stupidity is a stupid charge. -- Mike Temkin mst@csun.edu Cal. State U. Northridge, School of Engineering and Computer Science Voice phone: (818) 885-3919