RE: [PATCH] example and default IP addresses

["Dan Magenheimer" <dan.magenheimer@oracle.com>]

Hi Keir --

Per the below discussion, could you revert the change to network-nat
in the new cset 16739?

Thanks,
Dan

> -----Original Message-----
> From: xen-devel-bounces@lists.xensource.com
> [mailto:xen-devel-bounces@lists.xensource.com]On Behalf Of Ian Jackson
> Sent: Thursday, January 17, 2008 8:35 AM
> To: dan.magenheimer@oracle.com
> Cc: xen-devel@lists.xensource.com
> Subject: RE: [Xen-devel] [PATCH] example and default IP addresses
> 
> 
> Dan Magenheimer writes ("RE: [Xen-devel] [PATCH] example and 
> default IP addresses"):
> > In the patch to network-nat, I see that you are replacing 
> the 10.0.0.0/16
> > usage with 192.0.2.0/24.  Actually, vif-nat has a dependency on it
> > being 10.0.0.0/8(!), at least if more than 256 domains are 
> launched (not
> > necessarily simultaneously, just sequentially created and 
> destroyed).
> > In vif-nat ip_from_dom, IP's are created as 10.x.y.z for 
> vifw.z, where
> > x*256+y==w.
> 
> Firstly, I think it's important to note that network-nat and vif-nat
> are pretty ropey.  Anyone who is using them will almost certainly have
> had to adjust them to local conditions anyway.  For example, these
> scripts attempt to find and edit your local dhcp server configuration
> file !
> 
> > I'm not sure what the right answer is, but 192.0.2.0/24 definitely
> > doesn't have enough bits.  And regardless of the answer, 
> vif-nat will
> > need to be patched also.
> 
> Having said that, I think you're right.  vif-nat does indeed have to
> use 10/8 for static hosts for the reason you give.  Although really I
> think this is a poor approach.  So my change to network-nat ought not
> to be applied.
> 
> In practice many people using this code in any real setting are likely
> to fall foul of address clashes anyway as it's very likely that they
> have at least some use of 10/8 ...
> 
> Ian.
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel
>