Re: Bug#979432: ruby-rack FTBFS on IPV6-only buildds
- To: Adrian Bunk <bunk@debian.org>, 979432@bugs.debian.org
- Cc: debian-release@lists.debian.org
- Subject: Re: Bug#979432: ruby-rack FTBFS on IPV6-only buildds
- From: Chris Hofstaedtler <zeha@debian.org>
- Date: Sun, 7 Feb 2021 19:22:30 +0100
- Message-id: <[🔎] 20210207182229.3juupfluhraw6fei@percival.namespace.at>
- In-reply-to: <20210207181558.xj6updr32eqm45xt@percival.namespace.at>
- References: <160995375285.3256.15836480309689447233.reportbug@localhost> <20210207173018.ivpsgogaf54g5ial@percival.namespace.at> <160995375285.3256.15836480309689447233.reportbug@localhost> <20210207174634.GE32017@localhost> <160995375285.3256.15836480309689447233.reportbug@localhost> <20210207181558.xj6updr32eqm45xt@percival.namespace.at>
Control: found -1 2.1.4-1
Control: tags -1 - unreproducible
* Chris Hofstaedtler <zeha@debian.org> [210207 18:19]:
> Thanks. However that info appears outdated, as on a current Debian
> unstable, I always get a `lo` interface with 127.0.0.1 and ::1
> bound. The tests want 127.0.0.1, so that works.
>
> Without more info on how the buildd setup looks like, I cannot
> reproduce this. Or maybe the failure has gone anyway?
So, if I actually do this:
ip a del 127.0.0.1/8 dev lo
Then the tests indeed crash.
But now that modern kernels apparently always hand one a working
"127.0.0.1", do we really want to bother with this?
I guess one can change all the tests to use ::1 instead of
127.0.0.1, but that will just introduce other failure modes.
Chris
Reply to: