we (ietf 55 noc folks) have _indirectly_ heard complaints about
connectivity. the rumors we have heard are as follows:
o nokia APs crash under load "every 20 minutes" and nokia refused to use
cisco APs.
the APs are not nokia APs (the nokia sticker indicates to whom they
belong). the APs are lucent/proxim/whatever_they_are_calling_themselves_
this_week 2000s. nokia wanted to use the lucent APs and the lucent APs
have been used at previous meetings, but as we've discovered are far from
free of issues.
on advice from proxim DE and CE, various fixes have been tried and none
have solved all the problems.
cisco APs were shipped as backup and are helping to balance the load.
o when an AP dies, windows nt automatically switches to adhoc mode &
takes everyone else with them
windows xp and 2000 appear to have this problem.
o local and remote attacks on the ietf network
we have seen port scans, both internally and remotely.
o misconfigured APs with both radios on the same channel
true to an extent. this was done at proxim's request, regardless of our
insisting that it would cause collisions. the second proxim person agreed
with us and configs were updated.
o some wireless hosts not correctly configured
true. none of these seem to have been malicious; rather all have been
operating system problems. we've had some volunteer rogue APs.
o dhcp relay bug on junipers
there was a problem with dhcp relay between windows xp clients. we have
not determined if this was a problem with the juniper dhcp relay, the
server, or the client.
o some odd bug with interation with dhcp, dynamic dns, juniper where
juniper ends up arp'ing for every ip addr avail, and does not cache
no answer for a bit (may be same bug as above)
there is an excessive amount of arp'ing from the router. it is definitely
not related to dhcp or dynamic dns, but has been occuring since last
wednesday. we have not determined the cause, but it does not appear to
affect connectivity.
o some rumors of internal verio problems
false. the only external connectivity problems we are aware of have been
multicast related. we had problems with a switch and building cat5 run
in/to the salon control room there were loss points beyond verio borders
with peers.
excepting the window xp dhcp problem, we were aware of all the issues noted
above, but we are not psychic. folks have to tell the noc about observed
problems, rather than just complaining to your couch neighbor. the noc contact
information has been on the "network & terminal room" page link from the "host
infomation" page (http://www.ietf.org/meetings/IETF-55.html) since last
wednesday.
"How to get Network and Terminal Room Help
Contact the helpdesk open from 8AM to 8PM in the Sydney/Zurich/London/Bonn
terminal room OR send e-mail to rt(_at_)ops(_dot_)ietf(_dot_)org"