bug-inetutils
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Detecting IPv4 and IPv6 at runtime in Hydra builds


From: Mats Erik Andersson
Subject: Detecting IPv4 and IPv6 at runtime in Hydra builds
Date: Fri, 24 Jan 2020 19:23:55 +0100
User-agent: Mutt/1.5.23 (2014-03-12)

Hello there,

the last few days I have implemented runtime selection of IPv4 and IPv6
for the testing phase of Gnu Inetutils. The fact that a regular Raspbian
is set up without IPv6 surfaced when it gave false negatives for our
release candidate 1.9.5-RC1. On the other hand, I know that some effort
is invested into producing IPv6-only flavours of FreeBSD. Thus I under-
took the labour to craft said distinction at runtime.

The Hydra builds stumble severely in this new setting. The new test
program `runtime-ipv6' tries to resolve either `::1', or `127.0.0.1'
with a minimal call to getaddrinfo(). The first addition checked for IPv6
only and Hydra promptly failed [https://hydra.nixos.org/build/110721125/log]:

  IPv6 is disabled in this running system!

Today I pushed a larger set of changes to fully separate IPv4 from IPv6,
and to make them independently selectable. To my disappointment Hydra
now fails on both [https://hydra.nixos.org/build/110842965/log]:

   IPv4 is disabled in this running system: "127.0.0.1" Name or service not 
known
   Suppressing non-supported IPv4.
   IPv6 is disabled in this running system: "::1" Name or service not known
   Suppressing non-supported IPv6.

In effect, this failure of Hydra means that it is performing no Inet
socket testing at all, only Unix socket testing as well as some non-network
test of the executables. A very annoying situation. Is some vital setup
feature missing in Hydra? Why? Almost all our executables rely on
getaddrinfo(), yet they tested successful in past years. Why not now?

On behalf of Gnu Inetutils,

   Mats Erik Andersson



reply via email to

[Prev in Thread] Current Thread [Next in Thread]