On 26/07/11 22:45, Andrew O. Shadoura wrote:
> I think the version from experimental should work without the patch,
> please test.

I downloaded the source for ifupdown_0.7~alpha5.1 from the experimental repository on 28 July. "make" completed with a lot of warnings, but on inspection I felt none were relevant to my particular problem. I successfully installed the executables on my ubuntu 11.04 system with "sudo make install" (I created a deb but it wouldn't install because of missing pre-requisites - thanks for the tip, Andrew).

As predicted, the new ifupdown did NOT explicitly define an interface for 127.0.1.1 (or even 127.0.0.1), so I was curious about whether it would resolve my problem.

I ran a series of tests using ssh and sshd on the same system, but with different target host names. They worked differently, but not in the same way as with my patch applied to the latest version from the ubuntu repository (ifupdown_0.6.10ubuntu4).

I carefully compared my /etc/hosts file with one on a clean install of natty (also only 2 days old) and found some differences. I created a new hosts file...

127.0.0.1 localhost  localhost.mydomain
127.0.1.1 myhostname myhostname.mydomain
# the following lines are desirable for IPv6 capable hosts
</snip>

... and ran my tests again...

ssh localhost connected with IPv4 to sshd on 127.0.0.1

ssh myhostname connected with IPv4 to sshd on 127.0.1.1

ssh myhostname.mydomain connected with IPv4 to sshd on 127.0.1.1

These are the results that I wanted to see. netstat shows these listening server sockets for sshd:

tcp 0 0 0.0.0.0:22 0.0.0.0:* LISTEN 32515/sshd tcp6 0 0 :::22 :::* LISTEN 32515/sshd

... in other words, sshd no longer explicitly needs to listen on either of the loopback interfaces - the generic "all IPv4 interfaces" 0.0.0.0 is good enough. (As expected, other hosts could successfully connect to sshd on the test system via its IPv4 interfaces because my DNS resolves the hostname to only the IPv4 addresses).

I am satisfied the new version of ifupdown resolves my bug. However, I would be interested in a brief explanation of why it works because the source module is very hard for the uninitiated to read.

In particular, I would like to try back-porting just the change for my bug to the latest ubuntu source as a patch. I couldn't even identify the relevant update from the changelog!

Thanks!

Brian



--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org

Reply via email to