brian@natinst.com (Brian H. Powell) (05/30/90)
Man, this is driving me crazy. We got a GatorBox Friday, and I waited until today to mess with it. We also got GatorShare, which maps AFP to NFS. We were real excited; it was supposed to solve all of our problems. We plugged it in, turned it on, configured everything just so, downloaded the software. Poof, every one of our SCO UNIX '386 machines on ethernet died. (They just froze until a hard reboot.) Several people with other sorts of machines (e.g., PC/NFS) complained of an intolerably slow network for several seconds. Even someone with an rs232 connection to a Sun workstation complained that the machine was suddenly intolerably slow. Perhaps because that Sun was a YP server, and a network gateway. This behavior is reproducible. A Sparcstation 330 reported the following error: May 29 08:56:41 eagle vmunix: le0: Receive: giant packet from ff:ff:ff:ff:ff:ff May 29 08:56:41 eagle vmunix: le0: Receive: STP in rmd cleared Cayman's response was "WOW!". They're still working on it. Any ideas what is going on here? This is getting kind of annoying. Has anybody seen anything like this before? I can provide lots of details if you want to help me solve the problem. Brian H. Powell, M/S 56-14 National Instruments Corp. brian@natinst.com 6504 Bridge Point Parkway uunet!cs.utexas.edu!natinst!brian Austin, Texas 78730-5039 AppleLink:NATINST (512) 338-9119
BSCHMIDT@BNR.CA (Ben Schmidt, B.T.) (05/31/90)
Brian H. Powell, National Instruments writes: > We got a GatorBox Friday... > We plugged it in, turned it on, configured everything just so, > downloaded the software. Poof, every one of our SCO UNIX '386 > machines on ethernet died. (They just froze until a hard reboot.) > Several people with other sorts of machines (e.g., PC/NFS) > complained of an intolerably slow network for several seconds. > Even someone with an rs232 connection to a Sun workstation > complained that the machine was suddenly intolerably slow. Perhaps > because that Sun was a YP server, and a network gateway. I can't explain what's going on, but then again, a lot of things happen on my own network, which I can't explain. Anyhow, I wanted to add that when we first looked at PC/NFS, its default was to use the old-style all zero broadcast address. That means it wouldn't work here, as its broadcasts would be ignored by the rest of our IP nodes. As well, it also means, that it could ARP for the host portion of the broadcast address, which could smoke out other non-compliant IP implementations on the network, not to mention create a lot of unnecessary broadcast traffic. Mind you, since presumably your PC/NFS has been working to date, maybe all your IP nodes (except the GatorBox?) agreed to use all zero's as a broadcast address? :^) regards, ben, Information Technology/BNR