[1281] in SIPB-AFS-requests
Theories about ronald-ann problems
daemon@ATHENA.MIT.EDU (ghudson@MIT.EDU)
Wed Mar 30 13:25:23 1994
From: ghudson@MIT.EDU
Date: Wed, 30 Mar 94 13:23:32 EST
To: sipb-afsreq@MIT.EDU, sipb-staff@MIT.EDU
My observations about the ronald-ann outage:
At some point, ronald-ann stopped exchanging packets with any machine
using Dec hardware. It probably also stopped exchanging packets with
anything off the subnet, but I wasn't able to test this before the
problem spontaneously fixed itself.
Now, I could be drawing a false correlation, but we were able to ping
ronald-ann from deathtongue, steve-dallas, and oliver, but not from
podge, charon, maze, or ge. (maze is off the subnet, but is a
Decstation. Likewise, ge is a vax.)
This is too bizarre to be a network problem (dt and podge are on the
same piece of wire, notably). This is too bizarre to be a problem
with ronald-ann's network configuration. I suspect weird interactions
between ICMP packets sent by a rogue machine on our subnet and BSD
4.3/Ultrix routing behavior.
Total outage time was less than two hours and was completely a product
of the network.
--GBH
