[altq 266] Re: Message for altq

[Xin Wang <xwang@ctr.columbia.edu>]

On Wed, 5 Jan 2000, Kenjiro Cho wrote:

> 
> Alexandru Pirvulescu wrote:
> > I use pptp/pptpd for tunnelling or ppp over udp/tcp. When I start ALTQ
> > the ppp/pptpd daemon processes (that one that run on the server) are core
> > dumped with the following error(s):
> 
> [snip]
> 
> >   Same error ("No buffer space available"). I looked at the sources of
> > ppp and pptpd and i saw that it cannot send the packet through the interface
> >   I modified the sources not to coredump, and wait 10ms and my system
> > is going into an very strange state (squid reports that are too many opened
> > filesystems)
> > 
> >   I use a 256Kbit line which is split into 4 classes, 2 for interactive traffic
> > and 2 for web/ftp on 2 realtek ethernet cards (RTL8029) on interfaces ed1 and 
> > ed2.
> > 
> >   Somtimes I get the same error from ping command or the named daemon when
> > I run ALTQ. There are not such a big problem, but my tunnel is because
> > the link drops in maximum 10 seconds from the altq start.
> 
> I don't know much about pptpd but it seems you run ip over pptp over
> ppp over ethernet...
> 
> The current ALTQ classifier checks only the most outer header and,
> apparently, this is not what you want.
> Probably, you misconfigured filters and that's the cause of errors
> (ENOBUFS). 
> 
> -Kenjiro
>