Subject: Re: anoncvs problem
To: None <jsarkes@tiac.net>
From: Andrew Gillham <gillham@vaultron.com>
List: current-users
Date: 12/02/2000 13:25:19
Joseph Sarkes writes:
> 
> Hmmm, that appears to have fixed things here, THANKS :)

No problem.  Hopefully it works.

> At least there is now data transfer happening to the server. It
> will take a while for the sup to happen though (via ppp phone connection)
> The thing is though, that cvs worked fine for me till a couple days
> ago. What is this, and when did it change/what does it do?

RFC1323 defines a mechanism for increasing the TCP window size beyond
16bits. (by adding a "scaling factor")  It also defines a new timestamp
option.  Both of these are designed to improve performance/reliability
over connections just like my satellite. :(  In reality it appears that
either the spec is broken, NetBSD's implementation is broken, or someone
else's implementation is broken.  Or something. :-)

Anyway, I don't know why this would suddenly crop up on your end.  On my
end it was the addition of a satellite link with a 1460 MTU and a 600ms
RTT.  If I disconnect my satellite connection (e.g. hangup the outbound
side, the modem) and dialin with NetBSD the way I used to, it works fine.
It could just be the fact that my box has to route via a Windows 98
machine now.

For you, there might have been a change in the route between your box
and the NetBSD server, through some "broken" machine.

I think there would be more people complaining about this if it
was an issue with the NetBSD server end of it.

-Andrew