page@ulowell.UUCP (Bob Page) (08/05/86)
Configuration: VAX 750, Ultrix V1.0, 6MB memory, 2 RA-81s, TU-80, Old Interlan (link-level) Ethernet board, 3 DMF-32s. Problem: Ethernet doesn't seem to fully communicate. Symptoms: Power failure/surge (down for about 2 secs) the other night, and when the machine came back up we lost most of our Ethernet on this machine (other machines are fine). However, rwho and ruptime work OK coming in; that is, I can type either and get a report of the whole network, as normal. The other nodes don't get messages from us, and the other utilities (r, ftp, telnet) don't work in or out of the machine. I've checked all the config files (hosts, services, protocols, etc); nothing has changed from the time it worked and the time it didn't. All daemons have been restarted, etc. I would suspect a bad Interlan board or transceiver except that other hosts can send rwho/ruptime information to us! Local loopback tests (mail, script, talk, phone, etc) work. netstat(1) acts as if there are no problems. Any suggestions? I don't have a network analyzer, nor sources, today. Thanks in advance. ..Bob -- UUCP: wanginst!ulowell!page Bob Page, U of Lowell CS Dept VOX: +1 617 452 5000 x2233 Lowell MA 01854 USA
chris@umcp-cs.UUCP (Chris Torek) (08/07/86)
In article <625@ulowell.UUCP> page@ulowell.UUCP (Bob Page) writes: >Power failure/surge (down for about 2 secs) the other night, and >when the machine came back up we lost most of our Ethernet on this >machine (other machines are fine). However, rwho and ruptime work >OK coming in [...]. The other nodes don't get messages from us, >and the other utilities (r, ftp, telnet) don't work in or out of >the machine. >I would suspect a bad Interlan board or transceiver except that >other hosts can send rwho/ruptime information to us! Suspect it anyway. Your board is receiving broadcast packets, and no others. Broadcast packets likely use a different hardware path than `regular' packets. It sounds as though the address matcher in the board has quit. You cannot send anything because all outgoing packets get hung up waiting for ARP replies, which are never seen because the address matcher is broken. -- In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 1516) UUCP: seismo!umcp-cs!chris CSNet: chris@umcp-cs ARPA: chris@mimsy.umd.edu