on my iBook i've had some interesting behaviours when connecting to some nokia access points at a couple of airports (denver notably) i've found that these problems go away when i drop the MTU on the wireless adaptor to 1490. apple in their infinite wisdom has not opened up the driver details for the airport card. there are some folks that have reverse engineered some of the interesting parts from the 802.11.framework. there's a link on the macstumbler page [1] to more information as well as a nice airport utility that has the information that you're looking for. the utility is called airport and it lets you get all sorts of dirt on your airport card. when last we saw our hero (Monday, Jun 10, 2002), Brent J. Nordquist was madly tapping out: > [Sorry for the late reply to this:] > > On Tue, 4 Jun 2002, Andy Warner <andyw at pobox.com> wrote: > > > Brent J. Nordquist wrote: > > > > > It just appears that packets that are too large get > > > (consistently) dropped, which means that TCP sessions that only > > > use small packets (such as an interactive SSH session) work > > > fine, but as soon as a TCP session uses a large packet, that > > > session is no longer usable (since the retries are dropped too). > > > Other existing and new sessions continue to work fine. By > > > limiting the MTU I no longer experience it at all. > > > > Does either the AP or the PC card have the RTS/CTS setting enabled > > ? If so, what's the packet size set to ? > > The (LinkSys WAP11 v2.2) AP has the "RTS Threshold" set to 2432 and > "Fragmentation Threshold" set to 2346. The config. web page shows > those as the maximums, and the default values. I haven't figured > out how to check those values on the iBook yet (I'm running Mac OS X > and it's an Airport card). > > > Why are you not using a 1500 byte (e.g. ethernet) mtu ? > > Because when I do, packets over 576 in size get dropped on the > floor. An MTU of 576 on both sides causes the problem to go away. > I have no idea why a standard 1500 MTU doesn't work, or why a 576 > MTU gets me around the problem (and I am still curious); I just know > empirically that it does. I also know the iBook doesn't have this > problem with the APs at work. > [1] - http://homepages.mac.com/macstumber/ -- steve ulrich sulrich at botwerks.org PGP: 8D0B 0EE9 E700 A6CF ABA7 AE5F 4FD4 07C9 133B FAFC