RE: Using 127.n.x.x/16 for namespace to host

["Elwin Stelzer Eliazer" <stelzere-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>]

Thank you for the suggestion. The link local address should be good enough
for now.

Cheers,
Elwin.

-----Original Message-----
From: Guenter Roeck [mailto:groeck-gvzKVTG1yJJBDgjK7y7TUQ@public.gmane.org] 
Sent: Monday, July 06, 2009 5:29 PM
To: Elwin Stelzer Eliazer
Cc: containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org
Subject: Re: Using 127.n.x.x/16 for namespace to host

Look for ipv4_is_loopback().

Another option might be to use link local addresses.

Guenter

On Mon, Jul 06, 2009 at 03:46:20PM -0700, Elwin Stelzer Eliazer wrote:
> Hi,
> 
> I am using an internal bridge for internal socket communication between
the
> host and various containers' applications.
> This is purely an internal bridge and no external traffic is expected in
> this.
> I can use a regular private ip address subnet for this, for example a
> 192.168.x.x/16 for this, and things are working.
> But since the applications at the container and host handle external
traffic
> as well, we are forced to exclude this internal subnet for external use.
> Using a 127.n.x.x/16 subnet is preferred, since that is not expected to be
> used externally.
> But this is requiring kernel code change, since 127.0.0.1/8 is programmed
> for loopback interface 'lo' in host.
> Towards this tried changing:
>      IFF_LOOPBACK 0x10 // in in.h, changed from 0x8 to 0x10
>      ifa->ifa_prefixlen = 16; // in devinet.c, changed from 8 to 16
> And still do not see them connected.
> 
> Any suggestions on this will be appreciated.
> 
> regards,
> Elwin.
> _______________________________________________
> Containers mailing list
> Containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org
> https://lists.linux-foundation.org/mailman/listinfo/containers