Subject: Re: pppoe problem
To: Christopher Vance <firstname.lastname@example.org>
From: der Mouse <mouse@Rodents.Montreal.QC.CA>
Date: 12/14/2001 03:24:53
>> However, whenever I try to update yellow or blue using cvsup, I've
>> had the connection terminated with a write error after about 200kB
>> of traffic.
> This sounds like a path mtu discovery blackhole.
Doesn't sound like that to me. IME, PMTU-D blackholes usually lose
immediately, not only after over 100kB of traffic.
Though I suppose this depends on cvsup: does it always die at exactly
the same place, and is that place always the first time it wants to
send a "large" (roughly, >1500 bytes) amount of data without flushing
and blocking for a reply from the other end?
If not, it probably isn't PMTU-D.
This does sound remarkably like a problem I've had for a logn time,
wherein a connection will come up and start transferring data fast, but
gradually slow down and stall, and eventually time out. It doesn't
always time out at the same place, and I've not investigated far enough
to say much more - but these symptoms "smell" more like that than like
PMTU-D blackhole trouble.
> Without MSS clamping you won't be able to fix this.
The right fix is - if it really is a PMTU-D blackhole - to get the
broken router/host fixed. MSS clamping is at best papering over the
real problem, and it depends on a number of shaky assumptions (which,
to be fair, it looks to me as though you'd get away with; it may be
good enough for an interim patch for your case).
/~\ The ASCII der Mouse
\ / Ribbon Campaign
X Against HTML email@example.com
/ \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B