Re: Fedora rawhide FTBFS status 2010-02-10 x86_64

[Jaroslav Skarvada]

Following the steps from deltarpm example
(http://fedoraproject.org/wiki/UnderstandingDSOLinkChange#Example_deltarpm) helps me to
reproduce the DSO issue in local mock.

Jaroslav


----- Original Message -----
From: "Ralf Corsepius" <rc040203(a)freenet.de&gt;
To: "Development discussions related to Fedora"
<devel(a)lists.fedoraproject.org&gt;
Sent: Monday, February 15, 2010 8:55:03 AM GMT +01:00 Amsterdam / Berlin / Bern / Rome /
Stockholm / Vienna
Subject: Re: Fedora rawhide FTBFS status 2010-02-10 x86_64

On 02/13/2010 04:50 PM, Matt Domsch wrote:
>  Fedora Fails To Build From Source Results for x86_64
>  using rawhide from 2010-02-10
>
>  This run includes the new default linker feature --no-add-needed.
>  This is responsible for 446 failures noted below.  See
>  http://fedoraproject.org/wiki/Features/ChangeInImplicitDSOLinking for
>  further details.
>
>
>  Full logs at http://linux.dell.com/files/fedora/FixBuildRequires/ 
>  SoQt-1.4.1-14.fc13 (build/make) corsepiu 
I can't reproduce this issue in a local mock[1] using "released 
rawhide". Scratch-building using Fedora's infrastructure, however 
exposes this issue.

This renders investigating this issue quite difficult and imposes 
problems in fixing it[2]


What is required to be able to locally reproduce Matt's results?
- Is "released rawhide" different from what Matt rsp. "Fedora" is
using?
- Is "released rawhide" outdated/Do I happen to be using outdated 
rawhide mirrors?
- Are there some special settings required to trigger these DSO issues?

Ralf

[1] Standard "fedora-rawhide-*" mock setup on FC12 without any modification.

[2] I meanwhile have a "hack" which lets this package build again, but 
without being able to reproduce the issues in a local mock, I hardly can 
verify my "hack"'s correctness (This breakdown could easily be a 
side-effect of other bugs).
-- 
devel mailing list
devel(a)lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/devel

Fedora rawhide FTBFS status 2010-02-10 x86_64

[Ralf Corsepius]

On 02/15/2010 11:00 AM, Jaroslav Skarvada wrote:
>  Following the steps from deltarpm example
> (http://fedoraproject.org/wiki/UnderstandingDSOLinkChange#Example_deltarpm) helps me to
> reproduce the DSO issue in local mock. 
Thanks.

Hmm, however this would mean Matt's build-tree diverges from "released 
rawhide" i.e. his build-results are not actual, real "rawhide FTBS"
results.

Ralf

P.S.: I asked Matt whether his set up would be divering from "vanilla 
rawhide" on PM before asking here and he replied "not to be recalling 
any customization" ... Makes me wonder what is going on here.

We definitely have a communication problem somewhere.

Fedora rawhide FTBFS status 2010-02-10 x86_64

[Matt Domsch]

On Mon, Feb 15, 2010 at 11:49:47AM +0100, Ralf Corsepius wrote:
>  On 02/15/2010 11:00 AM, Jaroslav Skarvada wrote:
>  > Following the steps from deltarpm example
>  > (http://fedoraproject.org/wiki/UnderstandingDSOLinkChange#Example_deltarpm)
>  > helps me to reproduce the DSO issue in local mock. 
This would have been necessary before the Feb 9, which is when the new
linker made it into rawhide.  I specifically started my run on Feb 10
after ensuring that the new linker was in rawhide, and that my copy of
the tree was up-to-date, including that linker.

-- 
Matt Domsch
Technology Strategist, Dell Office of the CTO
linux.dell.com & www.dell.com/linux

Fedora rawhide FTBFS status 2010-02-10 x86_64

[Ralf Corsepius]

On 02/15/2010 01:48 PM, Matt Domsch wrote:
>  On Mon, Feb 15, 2010 at 11:49:47AM +0100, Ralf Corsepius wrote:
> > On 02/15/2010 11:00 AM, Jaroslav Skarvada wrote:
> >> Following the steps from deltarpm example
> >> (http://fedoraproject.org/wiki/UnderstandingDSOLinkChange#Example_deltarpm)
> >> helps me to reproduce the DSO issue in local mock.
>
>  This would have been necessary before the Feb 9, which is when the new
>  linker made it into rawhide.  I specifically started my run on Feb 10
>  after ensuring that the new linker was in rawhide, and that my copy of
>  the tree was up-to-date, including that linker. 
Which version of binutils exactly
have you been using?


I noticed my rawhide mock caches contain
binutils-2.20.51.0.2-13.fc13.x86_64

while download.fedora.redhat.com carries
binutils-2.20.51.0.2-15.fc13.x86_64


For reasons I don't yet now, mock doesn't install this (latest) version, 
so I am ending up with the *13.binutils (likely a version predating the 
chains)

=> I am inclined to think something is broken with the repos, their 
mirrors or with mock.

Ralf

Fedora rawhide FTBFS status 2010-02-10 x86_64

[Ralf Corsepius]

On 02/15/2010 02:28 PM, Ralf Corsepius wrote:
>  On 02/15/2010 01:48 PM, Matt Domsch wrote:
> > On Mon, Feb 15, 2010 at 11:49:47AM +0100, Ralf Corsepius wrote:
> >> On 02/15/2010 11:00 AM, Jaroslav Skarvada wrote:
> >>> Following the steps from deltarpm example
> >>> (http://fedoraproject.org/wiki/UnderstandingDSOLinkChange#Example_deltarpm)
> >>> helps me to reproduce the DSO issue in local mock.
> >
> > This would have been necessary before the Feb 9, which is when the new
> > linker made it into rawhide.  I specifically started my run on Feb 10
> > after ensuring that the new linker was in rawhide, and that my copy of
> > the tree was up-to-date, including that linker.
>  Which version of binutils exactly have you been using?
>
>
>  I noticed my rawhide mock caches contain
>  binutils-2.20.51.0.2-13.fc13.x86_64
>
>  while download.fedora.redhat.com carries
>  binutils-2.20.51.0.2-15.fc13.x86_64
>
>
>  For reasons I don't yet now, mock doesn't install this (latest) version,
>  so I am ending up with the *13.binutils (likely a version predating the
>  chains)
>
>  =>  I am inclined to think something is broken with the repos, their
>  mirrors or with mock. 
Update:

Meanwhile I rebuilt my rawhide-mock from scatch (rm -rf /var/cache/mock) 
and observed this:

- mock traversed 45 rawhide mirrors it considered to be broken before 
finding one it considered "current".

- now my mock picked up binutils*-15.fc13

- I still can't locally reproduce the FTBS breakdown.

Ralf