RE: Conntrack PPTP broken in 2.4.22 ?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Thanks for the thorough explanation Craig, The info you sent

should be made part of the "help" section in the kernel configuration of
the GRE / PPTP modules.

Any idea if the PPTP conntrack module will make it into the mainstream
kernel ? 


One more (unrelated?) question : is it possible to disable connection
tracking on a per interface basis ?

- Enrico

> -----Original Message-----
> From: Philip Craig [mailto:philipc@xxxxxxxxxxxx] 
> Sent: lunedì 15 settembre 2003 2.58
> To: Enrico Demarin
> Cc: netfilter@xxxxxxxxxxxxxxxxxxx
> Subject: Re: Conntrack PPTP broken in 2.4.22 ?
> 
> 
> Enrico Demarin wrote:
> > I think at this point I miss the functionality of the 
> pptp_conntrack 
> > module ? When is it necessary to load it ?
> 
> The module basically performs three tasks.
> 
> 1. NAT of the callid
> 
> This ensures that the PPTP callid is unique per client/server 
> pair. This is only necessary when you have multiple clients 
> that are NATed to the same source address.  Without it, the 
> server will get confused.
> 
> 2. Tracking of gre connections
> 
> The gre connections are tracked as RELATED connections, which 
> makes it easy to add a rule to let them through the firewall.
> 
> 3. NAT of gre connections
> 
> This ensures incoming gre packets are forwarded to the 
> correct internal server/client.
> 
> Since you typically only have one internal server, this is 
> just a convenience which means you don't need to add a gre DNAT rule.
> 
> However this is necessary if you have multiple clients that 
> are NATed to the same source address, so that the firewall 
> knows which client to forward the gre packet to.
> 
> NAT of gre can also be helpful for a single client.  Without 
> it, packets from the server may be dropped if the client 
> hasn't sent a packet recently.  This is because you typically 
> don't have a DNAT rule to forward gre to the client, so the 
> client has to send the gre packet first to establish the 
> conntrack, and this conntrack will timeout after 30-180 
> seconds of inactivity.  You can workaround this problem by 
> configuring PPP to send LCP echoes to keep the conntrack active.
> 
> -- 
> Philip Craig - philipc@xxxxxxxxxxxx - http://www.SnapGear.com 
> SnapGear - Custom Embedded Solutions and Security Appliances
> 




[Index of Archives]     [Linux Netfilter Development]     [Linux Kernel Networking Development]     [Netem]     [Berkeley Packet Filter]     [Linux Kernel Development]     [Advanced Routing & Traffice Control]     [Bugtraq]

  Powered by Linux