george@swbatl.sbc.com (George Nincehelser 5-6544) (11/02/90)
I've run across a strange problem with our company's AppleTalk networks. I've got a guy installing a Shiva Fastpath 4 several states away who is having problems seeing devices within certain zones that are local to me. He sees the zone names, but he can't see any devices in them. We've used InterPoll to confirm this (scanning for severl minutes). We are running a Phase 1 network. Any ideas? -- / George D. Nincehelser \ uunet!swbatl!george \ / / Southwestern Bell Telephone \ Phone: (314) 235-6544 \ / / / Advanced Technology Laboratory \ Fax: (314) 235-5797 \ / / / /\ 1010 Pine, St. Louis, MO 63101 \ de asini umbra disceptare \
jln@casbah.acns.nwu.edu (John Norstad) (11/03/90)
In article <1990Nov2.001254.12845@swbatl.sbc.com> george@swbatl.sbc.com (George Nincehelser 5-6544) writes: > I've run across a strange problem with our company's AppleTalk networks. > I've got a guy installing a Shiva Fastpath 4 several states away who is > having problems seeing devices within certain zones that are local to me. > He sees the zone names, but he can't see any devices in them. We've used > InterPoll to confirm this (scanning for several minutes). > > We are running a Phase 1 network. Very interesting. I'm currently struggling with similar problems, only in my case all my FastPath 4 boxes work fine. Instead, I'm having this same problem with Shiva NetBridges, a TurboBridge, and IPTalk/atalkad-tunneled Gatorboxes! As an example, I'll discuss a Shiva NetBridge case. Here's a picture of part of our large campus AppleTalk Internet (we have about 40 nets and 27 zones total in our Internet). All of the nets in this picture are LocalTalk. (For best results, view this picture using a monospace font!) /------------------/ net A /-----------------/ net B | | Shiva NetBridge Hayes InterBridge | | /---------------------------------------------------------/ net C | Hayes InterBridge | /-----------------------------/ net D | Shiva FastPath 4 | ... IPTalk Tunnel | Shiva FastPath 4 | /---------------------/ net E All 27 zone names appear properly in the Chooser on Macs on net A, but nothing can be found inside some of the zones (in either the Chooser or with Interpol). I'll call such zones "empty." In particular, in the diagram above, the zones for nets B and D are empty, while the zones for nets C and E are normal (non-empty). It's interesting that the empty zones are exactly the ones whose nets are immediately "behind" a Hayes InterBridge from net A's point of view. Zones whose nets are immediately "behind" other kinds of routers are normal. This is true not only in the small picture I've presented above, but throughout our complete Internet - there are about 10 empty zones behind InterBridges total from the point of view of Macs on net A. It's also interesting that net D is empty, while net E is normal, since traffic from net A must pass through net D to reach net E. So net A can see "through" net D, even though is can't see "in" net D. I examined the routing and zone information tables inside both the NetBridge and the InterBridge connecting nets B and C. Both were perfectly normal - correct network numbers, next router fields, hop counts, status, and zone names for all of our zones and nets. So it appears that the routers are communicating with each other using RTMP and ZIP correctly. In particular, the NetBridge's routing table entry for net B correctly pointed to the InterBridge joining nets B and C, with a hop count of 1, status "good", and the proper zone name in the ZIT. Similarly, the InterBridge's routing table entry for net A correctly pointed to the NetBridge with a hop count of 1, status "good", and the proper zone name in the ZIT. The problem is not symmetric: All the other nets on campus can see devices inside net A just fine. We're using phase 1 also. No phase 2 is involved anywhere here. I checked, and as far as I can tell I have NOT enabled any zone "hiding" or "filtering" options in any of my routers, and all of my routers are configured to do phase 1 routing, not phase 2. There must be something going wrong with the NBP Broadcast Request and zone-wide NBP Lookup broadcast mechanisms used to locate devices in zones. I wish I had a LocalTalk protocol analyzer to investigate this problem in more detail, but I don't (we have one on order). It's interesting that we're having this same problem with two different new Shiva NetBridges on campus (ROM v2.01). A third older Shiva NetBridge (ROM v1.02) on campus in a similar configuration works just fine! My problems with Gatorboxes and the TurboBridge are similar but not identical. They also involve this "empty zone" phenomenon. Has anybody else ever experienced this kind of problem? What can cause it? It's a mystery to me right now. I have a natural suspicion that my problems are all related and may have a common cause, but I have no idea what that cause might be. It is also possible that there is no common cause, and I really have three separate problems. John Norstad Academic Computing and Network Services Northwestern University jln@casbah.acns.nwu.edu
josh@CAYMAN.COM (Josh Littlefield) (11/04/90)
>> > I've run across a strange problem with our company's AppleTalk networks. >> > I've got a guy installing a Shiva Fastpath 4 several states away who is >> > having problems seeing devices within certain zones that are local to me. >> > He sees the zone names, but he can't see any devices in them. We've used >> > InterPoll to confirm this (scanning for several minutes). >> > >> > We are running a Phase 1 network. >> >> Very interesting. I'm currently struggling with similar problems, only in >> my case all my FastPath 4 boxes work fine. Instead, I'm having this same >> problem with Shiva NetBridges, a TurboBridge, and IPTalk/atalkad-tunneled >> Gatorboxes! [...lines deleted...] >> There must be something going wrong with the NBP Broadcast Request and >> zone-wide NBP Lookup broadcast mechanisms used to locate devices in zones. Sounds like a you are suffering from a Phase 1/2 incompatibility. Although you are not running Phase 2 EtherTalk, your new NetBridge may be running Phase 2 LocalTalk. For LocalTalk, there are indeed changes associated with Phase 2. Mostly it is in the form of the RTMP packet, which can now describe extended networks on the AppleTalk internet, and which was designed to be backward compatible. However, one additionally change was the addition of a the NBP Forward Request packet. A router receiving an NBP Broadcast Request turns it into an NBP Forward Request (instead of the old-style NBP Lookup directed broadcast) and sends it on. The last router is supposed to turn it into an NBP Lookup broadcast. But if the last router is a Phase 1 Hayes Interbridge, it won't work correctly. The FwdReq packet, whose DDP destination addr is (net N, node 0) will be dropped. The fact that the older Shiva NetBridge has no trouble, but the newer ones do would reinforce this guess. As for the GatorBox, we will generate a Forward request if the outbound interface is Phase 2, and a directed NBP Lookup instead if the interface is Phase 1. Release 1.5.0 always set our LocalTalk interface to Phase 2. Our latest releases (1.5.1 and now 1.6.0) allow LocalTalk to be set to Phase 1. We consider atalkad tunnels to be Phase 1 and GatorBox point-to-point tunnels to be Phase 2. >> I wish I had a LocalTalk protocol analyzer to investigate this problem in >> more detail, but I don't (we have one on order). You may be interested in trying Watch. Watch 1.6.1 is available for anonymous FTP from cayman.cayman.com (192.31.222.1) in ~ftp/pub/Watch/watch.1.6.1.sit.hqx. Watch will capture packets on either LocalTalk or Ethernet (although it only supports a few Ethernet cards). It does a pretty good job of protocol decoding and may help analyze your problems. I use it instead of a LocalTalk sniffer and am pretty happy with it. It is unsupported, but freely available. Constructive criticism is always welcome. Hope this helps. -josh ------------------------------------------------------------------------------ Josh Littlefield Cayman Systems, Inc. Engineering University Park at MIT josh@cayman.com 26 Landsdowne Street (617) 494-1999 Cambridge, MA 02139 ------------------------------------------------------------------------------