ken@helix.nih.gov (Ken Weeks) (12/04/90)
Thanks to Brad & Conrad for their ideas - I guess I should have been more specific... I've installed the Comm toolbox, if that makes any difference. My Mac is on the Ether directly, via a Cabletron card. We do have a FastPath (my apologies, Brad!) doing KIP routing and dynamic IP number assignment for the LocalTalk folks, but I've got a static IP address - should be the simple set-up here. I tried varying the max segment size a bit, but got the same hangup. I tried re-installing the Cabletron driver software, but with no success. I mention the card because one of my colleagues down the hall *DOESN'T* get the hang up - he's got a Nuvotech card, but he's also on a IIfx with lots of other differences. Oh, and the LocalTalk folks don't hang up, either - but it's as reliable as sunrise on this box. It doesn't seem to be time-dependent - I can make it hang almost immediately. I have had a problem with a MacTCP/VMS Telnet not getting along before. My supplier, Process Software, sent me a simple patch, with a warning that "whoever" wrote that client software wasn't doing "proper PSH notification" (?). Once again, NCSA Telnet for the PC and the Mac version with the embedded driver seemed to have no problems (cursor keys wouldn't work), so I kinda suspected MacTCP. *BIG DISCLAIMER* - I have absolutely no idea just WHAT is going on! I know vendors love to point at the other guy and say "It's HIS fault", so I mention this just for reference. I'm posting this message using MacTCP. I really love it. Honest! (-: Anyway, the problem seems limited to this machine. At least my users aren't breathing down my neck for answers! I don't know if there's a general interest here. If you'll reply by E-mail, I'll summarize and post as soon as I figure this *&^%# thing out. Ken Weeks National Institutes of Health, Bethesda, MD. "Once in a while you get shown the light, In the strangest of places, if you look at it right..."
resnick@cogsci.uiuc.edu (Pete Resnick) (12/04/90)
ken@helix.nih.gov (Ken Weeks) writes: >My Mac is on the Ether directly, via a Cabletron card. OK, now I have a guess, and this will probably work. Evidently some Ethernet card makers don't deal properly with maximum segment size that MacTCP tells it to have, though somehow the non-MacTCP version of NCSA Telnet does not have this problem. (Of course, MacTCP does not use config.tel to figure out the segment size, as one person thought, but takes care of it itself). The behavior you describe is *exactly* the behavior of my Dove SE card when I am connected to a local (no routers) IBM RT, thought it does not happen on a Sun or an Ultrix machine, or when I am going through a router. I would not be suprised if Cabletron has the same problem. Luckily, the answer is to use the same patch that was used to fix the problem with Localtalk macs in MacTCP 1.0 with MacTCP 1.0.1. Steve Dorner at UIUC wrote the patch. Using ResEdit or equivalent, go into the MacTCP file and open the DRVR resource (in ResEdit 1.2, I think you had to choose Open General... to get it open). Then do an "Open using Hex editor" so that you can see the Hex code for the ".ipp" driver. Find the string "337c02040014" and replace it with the string "337c01010014". Then save and close the file. Evidently, this just tells MacTCP not to say anything about segment size, and that should do it. Good luck, 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
lloyd@aplcen.apl.jhu.edu (Lloyd W. Taylor) (12/04/90)
Two personal experiences which may be relevant: 1. I had a CableTron card in my IIx for awhile, but removed it when it had trouble with various services using MacTCP. I'm back to the Novell/Excelan/Kinetics EtherPort II now, and am not having problems. 2. I'm a beta tester for TCP/Connect II. For a while, we were having trouble using it to talk to 3270 devices. According to the author, the problem was with how MacTCP handled the "Push Bit". He developed a work around which solved the problem for TCP/Connect. -- Lloyd Taylor Network Manager "Everything on my Mac is beta!!"