Subject: Re: pppd problem
To: None <lainestump@rcn.com>
From: Brad Spencer <brad@anduin.eldar.org>
List: netbsd-help
Date: 02/24/2001 09:31:49
   "Kevin P. Neal" <kpneal@pobox.com> writes:

   > On Sat, Feb 24, 2001 at 07:31:25AM +0800, Clarence Chan wrote:
   > > Hello,
   > > 
   > > THe pppd works for me for a long time.  However,  I found out that there
   > > is a little problem of it.  The error is
   > > 
   > > ================error=============
   > > pc135# ping b1
   > > PING b1 (208.151.73.70): 56 data bytes
   > > ping: sendto: No buffer space available
   > > ping: sendto: No buffer space available
   > 
   > Well, either you ran out of buffer space or you hit the other resource
   > shortage (that I can't remember now). 

   I'm guessing it means that there is something "bad" about the
   connection which prevents the modem from sending any packets, so the
   send queue fills up.

It is also very possible that the link was overbusy.  I got this quite
often on my V.34 PPP link to the Net when it became overbusy.  Usually the
first thing to lose was ICMP, followed by UDP.  Of course, when UDP went,
so did things like DNS lookups.

   > I get nailed by this every so often. I keep meaning to write a script
   > to check for it and bounce my connection.

   Yep, when the line conditions degrade, that's about all you can do.

This didn't always help me if the desire for the bandwidth never lessened,
after the link was reestablished.

   Of course, if LQM (Line Quality Monitoring) is working properly, pppd
   should notice this condition itself and cycle the line. Have you tried
   watching tcpdump on your ppp interface to see if LQM messages are
   being sent/received?

I was not aware that PPP had such a thing.  What criteria does it use to
determine "quality"??



Brad Spencer - brad@anduin.eldar.org
http://anduin.eldar.org  - & -  http://mellon.ipv6.eldar.org [IPv6 only]
[finger brad@anduin.eldar.org for PGP public key]