[dns-operations] The Python 3 socket.getaddrinfo() Case Sensitivity Problem

[m3047]

I'll skip the lament about how once upon a time we set these things 
ourselves and knew what they were... If it does come back it'll be good to 
have these notes.

On Fri, 5 Jul 2019, Florian Weimer wrote:
>> TLDR: I applied some OS updates and now can't reproduce the problem. Sorry
>> to be saying this, not sorry to see the problem gone. If it comes
>> back, I'll spend some more time tracking it down.

The OS update which presumably "fixed" it occurred on July 3rd or 4th.

>> [...]
>
> Which operating system is this?

Various flavors of SuSE Linux both across different machines and over 
time. The box which was having the problem (10.0.0.224) was ironically the 
one hosting Redis. For comparison I'm providing the machine hosting 
the caching resolver 10.0.0.220.

10.0.0.220 was newly built with Leap 15.0 the last week of December 2018.
10.0.0.224 was newly built with Leap 15.0 the last week of January 2019.

Neither machine uses DHCP.

> Has it NSS?

Yes.

> What does the hosts line
> in /etc/nsswitch.conf say?

   hosts:          files mdns_minimal [NOTFOUND=return] dns
   networks:       files dns

It's been that way since 2015 according to backups, although that is prior 
to Leap 15.0 (same hostname, different hardware).

Multicast DNS is not used here.

Prior to the last week of January the hosts: entry for 10.0.0.220 was:

   hosts:          files dns

although since then it has been:

   hosts:          files mdns_minimal [NOTFOUND=return] dns

Recall that this box was built at the end of December; I don't recall 
explicitly changing it.

While I'm at it here's /etc/resolv.conf:

Prior to upgrading to 15.0 10.0.0.224 had:

   search m3047
   nameserver 10.0.0.220

After upgrading to 15.0:

   search m3047 m3047.net
   nameserver 10.0.0.220

For 10.0.0.220 prior to January:

   search m3047.net zipcon.net
   nameserver 127.0.0.1

and since then:

   search m3047
   nameserver 127.0.0.1

Regarding /etc/hosts, prior to January 10.0.0.224 had:

   10.0.0.224      sophia.m3047 sophia

and since then, up until June:

   10.0.0.224    sophia

and since June:

   10.0.0.224    sophia
   10.0.0.224    sophia

Yes, that's right: since June (backup on the 22nd) there have apparently 
been two identical entries in there. I don't know why. It's being managed 
by wicked.

Prior to January 10.0.0.220 had:

   10.0.0.220          athena.m3047.net        athena
   209.221.140.128     athena.m3047.net

and since then:

   10.0.0.220          athena
   209.221.140.128     athena

with no change in June.

--

Fred Morris