Eugene.Hastings@MORGUL.PSC.EDU (06/13/88)
We have encountered a peculiar situation on a Proteon p4200 that has broken connectivitiy for Ohio State, Case Western Reserve, and the Universith of Michigan. The router in question is located in Cleveland and has T1 lines to those institutions, and also to the routing stub where PSC-GW and companions connect the ARPANET, NSFNET, PSCNET, and SURANET. Explicit RIP routes are propagated, plus translated (from hello) NSFNET routes, plus default. (Let me remind all and sundry that the present gated does not announce routes learned via EGP, but relies on the propagation of default.) So far, so good. Default is propagated successfully to the affected schools, but when data is sent to a net via default, the p4200 in Cleveland returns net unreachable (in fact, if the Proteon console is queried for a route to such a net, like MILNET, the response is "unable to route, reason none". This reouter has been running v7.4 of Proteon's code for some time, and it has only been very recently that it has failed. Proteon's response has been to advise us to use a bugfix, v7.4b, which we have only just received. No explanation of what may have happened to exercise this bug, or precisely what bugs have been fixed has been offered. The new version should be installed today. Gene
Eugene.Hastings@MORGUL.PSC.EDU (06/14/88)
Yea, verily, and after a restart, the Router In Question passes previously barfed traffic. Thanks, and a tip of the bashed hat. Yet another reminder that confession can be good for one's productivity, let alone one's soul. Thanks again, Gene - - - - Begin forwarded message - - - - Date: Mon, 13 Jun 88 08:25:13 PDT From: dlw@violet.Berkeley.EDU (David Wasley) Message-Id: <8806131525.AA24652@violet.berkeley.edu> To: Eugene.Hastings@morgul.psc.edu Subject: p4200 bug We (actually Milo Medin) found what may be that bug when we installed 7.4 here, and Proteon generated 7.4b for us as a result. Basically, if the box had an explicit route for a net, and it became unreachable (ie 16), the box wouldn't take default but would claim the net was explicitly unreachable. There is a mailing list for p4200 users that Scott Brim started long ago: p4200@devvax.tn.cornell.edu -- some of those folks may have other bug information re 7.4. We've been using 7.4b since January. David Wasley U C Berkeley - - - - End forwarded message - - - -
evan@ndcheg.UUCP (Evan Bauman) (06/16/88)
In article <1988.6.13.14.27.28.Eugene.Hastings@morgul>, Eugene.Hastings@MORGUL.PSC.EDU writes: > We have encountered a peculiar situation on a Proteon p4200 that has broken > connectivitiy for Ohio State, Case Western Reserve, and the Universith of > Michigan. The router in question is located in Cleveland and has T1 lines to > those institutions, and also to the routing stub where PSC-GW and companions > connect the ARPANET, NSFNET, PSCNET, and SURANET. Explicit RIP routes are > propagated, plus translated (from hello) NSFNET routes, plus default. (Let > me remind all and sundry that the present gated does not announce routes > learned via EGP, but relies on the propagation of default.) > > So far, so good. Default is propagated successfully to the affected schools, > but when data is sent to a net via default, the p4200 in Cleveland returns > net unreachable (in fact, if the Proteon console is queried for a route to > such a net, like MILNET, the response is "unable to route, reason none". > This reouter has been running v7.4 of Proteon's code for some time, and it > has only been very recently that it has failed. Proteon's response has been > to advise us to use a bugfix, v7.4b, which we have only just received. > No explanation of what may have happened to exercise this bug, or precisely > what bugs have been fixed has been offered. The new version should be > installed today. Just wanted to throw my two cents in on the subject. Although I'm posting from Notre Dame, I'm currently residing at the University of Cincinnati. They guys running the HPUX box here are new to the news and nets and have compiled the news a la FASCIST. So it's time to telnet back home to post until they work out outbound news with the upstream guys. In any case, I believe that we here are part of the Ohio network that is having its share of troubles. We've been experiencing the same "network not reachable" messages here. I've recently configured a set of Suns to work with a Proteon p4200, and from my experience there, I can tell that the routing tables on the HPUX here is correct. In fact, there were times that I was able to reach several sites that were not explicitly listed in the netstat -r output. Then sometime on Monday, net access was perfect. I could reach every site I could think of. This would coincide with the announcement from OSU that the new Proteon gateway software was installed. On Monday night, the net failed again here in Cincinnati. We got mostly "network not reachable" messages, but for a while, the attempted connection just timed out. Suddenly, tonight it came back up; as witnessed by my ability to post in Indiana while sitting at a desk in Ohio. I don't mean to gripe. Was just trying to let the guys in Columbus know what's happening here and help with the diagnostics. Evan Bauman University of Notre Dame ..!iuvax!ndcheg!evan <-- will forward to proper U. Cin. address -- --------------------------------------------------------------------------- This message has been sponsored by Powdermilk Biscuits in the big blue box. ---------------------------------------------------------------------------