From: Jacques Gelinas (jack_at_solucorp.qc.ca)
Date: Tue 13 May 2003 - 23:19:23 BST
On Thu, 8 May 2003 21:53:05 -0500, Keith Smethers wrote
> The test is section 2.6.3, "Playing with /usr/sbin/chbind", from "Introduction to
> vserver" on the
> vserver homepage. If you confirm my results, comparing with ctx16 would be a debugging
> start as
> Georges Toth responded saying he successfully tested chbind in ctx16. Below is my
> post describing
> the problem.
Ok, got it!
Indeed we did a change of semantic lately. The idea is that if you try to connect to
127.0.0.1, you source address will be 127.0.0.1 as well. The ipv4root is forced
on outgoing connection only for other IP.
This was done because some software have a single way to accept connection
(IP socket), and test 127.0.0.1 specifically to tell if this is a local connection.
Now, the documentation should be reworked.
Did you experienced problems around this stuff or you were only testing ?
Note that at some point, vservers will have a real private loopback address. Now
that they can listen to all IPs they have at once (bind any works), we can address
this. This will allow a service to bind to 127.0.0.1 and other IP using different sockets.
---------------------------------------------------------
Jacques Gelinas <jack_at_solucorp.qc.ca>
vserver: run general purpose virtual servers on one box, full speed!
http://www.solucorp.qc.ca/miscprj/s_context.hc