Bug 10254

Created attachment 4041 [details]
urpmi.cfg

Is this a new install ? if yes what for media did you use ?

Keywords: (none) => NEEDINFO
Source RPM: (none) => rpmdrake

(In reply to Manuel Hiebel from comment #2)
> Is this a new install ? if yes what for media did you use ?

Yes, this is a new install from Mageia-3-x86_64-DVD.iso

Keywords: NEEDINFO => (none)

then it is strange was nobody else noticed that on all qa test (or nobody else in the bugzilla)

Status: NEW => UNCONFIRMED
Component: Release (media or process) => RPM Packages
Ever confirmed: 1 => 0

I can confirm. I did a fresh install from the same media (64bit DVD) and all nonfree repos are selected.

CC: (none) => philippe.l

Fixed in git

Status: ASSIGNED => RESOLVED
Resolution: (none) => FIXED

Please let in this urpmi update for mga2

Advisory:
==========
This update of urpmi fixes a bug where extra tainted and/or nonfree media (Backports*, Updates Testing) would be enabled if  tainted and/or nonfree packages would be installed (mga#10254).

Status: RESOLVED => REOPENED
Resolution: FIXED => (none)
Assignee: bugsquad => qa-bugs

packages are *urpmi*7.22.2*

I have updated urpmi.

I have removed medias with urpmi.removemedia -a

I added medias with the GUI => Same problem.

As of urpmi-7.22.2, I cannot reproduce what you say.
I only got the 4 (Tainted|NonFree) (Release|Updates) that got enabled

Bug confirmed in a new install of Mageia 3, where I
added the media after installation.

urpmq --list-media active
Core Release
Core Updates
Nonfree Backports
Nonfree Backports Testing
Nonfree Release
Nonfree Updates
Nonfree Updates Testing

I'll attach more output.

CC: (none) => davidwhodgins

We don't need bug confirmation anymore since it's already fixed

Created attachment 4048 [details]
Output of urpmi.addmedia --debug

During all prior testing of Mageia 3, I was always able to add my local
mirror using
urpmi.addmedia --distrib http://192.168.10.101/distrib/3/i586
even though my local repo doesn't have the srpms or debug repos.

Granted, I was using cauldron, not 3.

Now it's failing with no metadata found for medium "Nonfree Backports Debug"

So I temporarily edited distrib/3/i586/media/media_info/media.cfg on my
local repo, removing the srpm  and debug repos, after which I was able to
add the repos, but then found all existing Nonfree repos were enabled.

So something is triggering trying to activate media that shouldn't be activated,
which, if not present, is preventing the addition of the repos.

This was a new install (lxde) using the i586 dvd.

Note that I never saw this during qa testing of the iso images, and currently
have no idea what is triggering it.

After installing the urpmi update, removing all media, and readding
my local repo ...

urpmq --list-media active
Core Release
Core Updates
Nonfree Release
Nonfree Updates

Testing complete on Magiea 3 i586.

I'll test x86_64 shortly.

Testing complete on Mageia 3 x86_64

Could someone from the sysadmin team push the srpm
urpmi-7.27.2-1.mga3.src.rpm
from Mageia 3 Core Updates Testing to Core Updates.

Advisory: This update of urpmi fixes a bug where extra tainted and/or nonfree media (Backports*, Updates Testing) would be enabled if  tainted and/or nonfree packages would be installed (mga#10254).

https://bugs.mageia.org/show_bug.cgi?id=10254

Keywords: (none) => validated_update
Hardware: x86_64 => All
Whiteboard: (none) => MGA3-64-OK MGA3-32-OK
CC: (none) => sysadmin-bugs

Oh, this one can bite us hard for all of mga3 lifetime :/

Update pushed:
https://wiki.mageia.org/en/Support/Advisories/MGAA-2013-0022

Status: REOPENED => RESOLVED
CC: (none) => tmb
Resolution: (none) => FIXED

I wonder if this update would have been better if it in %post would have disabled all the:

Backports
Backports Testing
Updates Testing

medias to save all users that dont check what happends when they install an update...

Yeah I though about that too, but after pushing this.
Should a %triggerpost though.
We can still push another one...

I think we should do it, or we will end up with a lot of endusers getting packages installed that normally only QA or advanced users installs...
and that will flood us with bugreports...

I do realize that some wants the testing & other medias enabled, and this will make some of them have to enable those medias again, but I think thats the better alternative.

Yes, it looks like a good thing to do.

CC: (none) => boklm

*** Bug 10302 has been marked as a duplicate of this bug. ***

CC: (none) => isolde

(In reply to Thomas Backlund from comment #19)
> I think we should do it, or we will end up with a lot of endusers getting
> packages installed that normally only QA or advanced users installs...
> and that will flood us with bugreports...
> 
> I do realize that some wants the testing & other medias enabled, and this
> will make some of them have to enable those medias again, but I think thats
> the better alternative.

If we do, please include a README.upgrade.urpmi,
explaining that they are being disabled, to
ensure anyone who does want them enabled will
know to re-enable them.

(In reply to Dave Hodgins from comment #22)
> If we do, please include a README.upgrade.urpmi,
> explaining that they are being disabled, to
> ensure anyone who does want them enabled will
> know to re-enable them.

I don't think such README is needed. Most users don't need such repos and wouldn't enable them anyway. This README is just confusing for them. We should do this as one-time-operation only. Disable medias that we enabled and touch some marker file to check on next urpmi update to not do it again if user has enabled some medias again.

CC: (none) => sander.lepik