FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Redhat > Fedora Build System

 
 
LinkBack Thread Tools
 
Old 03-06-2010, 01:16 PM
Jeroen van Meeuwen
 
Default mock 1.0.5 failure

>From the mock-1.0.5 RPM changelog, I read:

==
* Wed Feb 17 2010 Clark Williams <williams@redhat.com>- 1.0.5-1
- from Jesse Keating <jkeating@redhat.com>:
- fixed 'useradd' option conflict with EPEL (-N vs -n)
- added Fedora 13 configs
==

>From the build logs of an Enterprise Linux 5 build (executed on a Fedora
12 koji build host) I read:

==
DEBUG util.py:280: Executing command: /usr/sbin/useradd -o -m -u 490 -g
490 -d /builddir -N mockbuild
DEBUG util.py:256: /usr/sbin/useradd: invalid option -- N
DEBUG util.py:256: Usage: useradd [options] LOGIN
DEBUG util.py:256: Options:
DEBUG util.py:256: -b, --base-dir BASE_DIR base directory for the new
user account
DEBUG util.py:256: home directory
DEBUG util.py:256: -c, --comment COMMENT set the GECOS field for the
new user account
DEBUG util.py:256: -d, --home-dir HOME_DIR home directory for the new
user account
DEBUG util.py:256: -D, --defaults print or save modified default useradd
DEBUG util.py:256: configuration
DEBUG util.py:256: -e, --expiredate EXPIRE_DATE set account
expiration date to EXPIRE_DATE
DEBUG util.py:256: -f, --inactive INACTIVE set password inactive
after expiration
DEBUG util.py:256: to INACTIVE
DEBUG util.py:256: -g, --gid GROUP force use GROUP for the new user
account
DEBUG util.py:256: -G, --groups GROUPS list of supplementary groups
for the new
DEBUG util.py:256: user account
DEBUG util.py:256: -h, --help display this help message and exit
DEBUG util.py:256: -k, --skel SKEL_DIR specify an alternative skel
directory
DEBUG util.py:256: -K, --key KEY=VALUE overrides /etc/login.defs
defaults
DEBUG util.py:256: -m, --create-home create home directory for the
new user
DEBUG util.py:256: account
DEBUG util.py:256: -l, do not add user to
lastlog database file
DEBUG util.py:256: -M, do not create user's
home directory(overrides /etc/login.defs)
DEBUG util.py:256: -r, create system account
DEBUG util.py:256: -o, --non-unique allow create user with duplicate
DEBUG util.py:256: (non-unique) UID
DEBUG util.py:256: -p, --password PASSWORD use encrypted password for
the new user
DEBUG util.py:256: account
DEBUG util.py:256: -s, --shell SHELL the login shell for the new
user account
DEBUG util.py:256: -u, --uid UID force use the UID for the
new user account
DEBUG util.py:256: -Z, --selinux-user SEUSER use a specific SEUSER
for the SELinux user mapping
DEBUG util.py:319: Child returncode was: 2
DEBUG util.py:98: kill orphans
==

Has the fixed broken builds on EL5? Should any changes be applied on the
consuming edge, and if yes, why was this released to a stable Fedora
branch??

-- Jeroen
--
buildsys mailing list
buildsys@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/buildsys
 
Old 03-06-2010, 01:49 PM
Till Maas
 
Default mock 1.0.5 failure

On Sat, Mar 06, 2010 at 03:16:44PM +0100, Jeroen van Meeuwen wrote:

> Has the fixed broken builds on EL5? Should any changes be applied on the
> consuming edge, and if yes, why was this released to a stable Fedora
> branch??

You need to merge the new EPEL mock config, it contains some extra
option to fix this:
config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s -d
%(home)s -n %(user)s'

I cannot tell you why it has been pushed, probably the -1 karma came to
late:
https://admin.fedoraproject.org/updates/mock-1.0.5-1.fc12

Regards
Till
--
buildsys mailing list
buildsys@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/buildsys
 
Old 03-06-2010, 03:58 PM
Jeroen van Meeuwen
 
Default mock 1.0.5 failure

On 03/06/2010 03:49 PM, Till Maas wrote:
> On Sat, Mar 06, 2010 at 03:16:44PM +0100, Jeroen van Meeuwen wrote:
>
>> Has the fixed broken builds on EL5? Should any changes be applied on the
>> consuming edge, and if yes, why was this released to a stable Fedora
>> branch??
>
> You need to merge the new EPEL mock config, it contains some extra
> option to fix this:
> config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s -d
> %(home)s -n %(user)s'
>

The builds I'm talking about are executed through Koji; The only thing I
can find is that /etc/mock/site-defaults.cfg has:

# config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s
-d %(home)s -n %(user)s' # Fedora/RedHat

whereas /etc/mock/site-defaults.cfg.rpmnew has:

# config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s
-d %(home)s -N %(user)s' # Fedora/RedHat

So I suppose the default changed and is used for EL-5 builds.

Given that my Fedora builders build EL-5 packages as well, how would I
fix this? In Koji or in mock?

-- Jeroen
--
buildsys mailing list
buildsys@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/buildsys
 
Old 03-06-2010, 04:11 PM
Todd Zullinger
 
Default mock 1.0.5 failure

Jeroen van Meeuwen wrote:
> The builds I'm talking about are executed through Koji; The only thing I
> can find is that /etc/mock/site-defaults.cfg has:
>
> # config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s
> -d %(home)s -n %(user)s' # Fedora/RedHat
>
> whereas /etc/mock/site-defaults.cfg.rpmnew has:
>
> # config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s
> -d %(home)s -N %(user)s' # Fedora/RedHat

As an aside, the comment "Fedora/RedHat" is now untrue. It should
read "Fedora/RedHat > 6" perhaps.

> So I suppose the default changed and is used for EL-5 builds.
>
> Given that my Fedora builders build EL-5 packages as well, how would I
> fix this? In Koji or in mock?

You can override the defaults from site-defaults in your epel config
files. Adding:

config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s -d %(home)s -n %(user)s'

to epel-{4,5}*.cfg should make mock use the right command for EL
builds.

--
Todd OpenPGP -> KeyID: 0xBEAF0CE3 | URL: www.pobox.com/~tmz/pgp
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~~
Let us think the unthinkable, let us do the undoable. Let us prepare
to grapple with the ineffable itself, and see if we may not eff it
after all.
-- Douglas Adams

--
buildsys mailing list
buildsys@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/buildsys
 
Old 03-06-2010, 04:47 PM
Jeroen van Meeuwen
 
Default mock 1.0.5 failure

On 03/06/2010 06:11 PM, Todd Zullinger wrote:
> Jeroen van Meeuwen wrote:
>> The builds I'm talking about are executed through Koji; The only thing I
>> can find is that /etc/mock/site-defaults.cfg has:
>>
>> # config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s
>> -d %(home)s -n %(user)s' # Fedora/RedHat
>>
>> whereas /etc/mock/site-defaults.cfg.rpmnew has:
>>
>> # config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s
>> -d %(home)s -N %(user)s' # Fedora/RedHat
>
> As an aside, the comment "Fedora/RedHat" is now untrue. It should
> read "Fedora/RedHat > 6" perhaps.
>
>> So I suppose the default changed and is used for EL-5 builds.
>>
>> Given that my Fedora builders build EL-5 packages as well, how would I
>> fix this? In Koji or in mock?
>
> You can override the defaults from site-defaults in your epel config
> files. Adding:
>
> config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s -d %(home)s -n %(user)s'
>
> to epel-{4,5}*.cfg should make mock use the right command for EL
> builds.
>

Koji writes out it's own mock configuration files though. As a result,
/etc/mock/koji/*.cfg (per buildroot) uses the mock default -n/-N. Either
one is incompatible with the other, and I have not found an alternative.

No setting in any epel*-*.cfg in /etc/mock/. can change that wrt. to
what Koji executes. Instead, I've removed the -n/-N in
/etc/mock/site-defaults.cfg

-- Jeroen

--
buildsys mailing list
buildsys@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/buildsys
 
Old 03-08-2010, 08:09 PM
Jesse Keating
 
Default mock 1.0.5 failure

On Sat, 2010-03-06 at 15:16 +0100, Jeroen van Meeuwen wrote:
> Has the fixed broken builds on EL5? Should any changes be applied on the
> consuming edge, and if yes, why was this released to a stable Fedora
> branch??

Fun. Looks like we forgot to test this in a koji environment, since
koji creates those configs on the fly. We didn't run into this with
Fedora's koji as we manage when updates get applied there.

The problem is that the previous default was wrong for anything newer
than EL5, so we changed the default, and overrode it in the EPEL
configs. Not sure what the best thing to do going forward would be,
I'll have to look at the koji package and see if we can change it there
too.

--
Jesse Keating
Fedora -- Freedom˛ is a feature!
identi.ca: http://identi.ca/jkeating
--
buildsys mailing list
buildsys@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/buildsys
 
Old 03-09-2010, 08:50 PM
Jeroen van Meeuwen
 
Default mock 1.0.5 failure

On 03/08/2010 10:09 PM, Jesse Keating wrote:
> On Sat, 2010-03-06 at 15:16 +0100, Jeroen van Meeuwen wrote:
>> Has the fixed broken builds on EL5? Should any changes be applied on the
>> consuming edge, and if yes, why was this released to a stable Fedora
>> branch??
>
> Fun. Looks like we forgot to test this in a koji environment, since
> koji creates those configs on the fly. We didn't run into this with
> Fedora's koji as we manage when updates get applied there.
>
> The problem is that the previous default was wrong for anything newer
> than EL5, so we changed the default, and overrode it in the EPEL
> configs. Not sure what the best thing to do going forward would be,
> I'll have to look at the koji package and see if we can change it there
> too.
>

Since the -n/-N/--no-user-group option only controls whether a user
specific/primary group should be created, can't we just drop it?

-- Jeroen
--
buildsys mailing list
buildsys@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/buildsys
 
Old 03-11-2010, 10:55 PM
Jesse Keating
 
Default mock 1.0.5 failure

On Sat, 2010-03-06 at 18:47 +0100, Jeroen van Meeuwen wrote:
> On 03/06/2010 06:11 PM, Todd Zullinger wrote:
> > Jeroen van Meeuwen wrote:
> >> The builds I'm talking about are executed through Koji; The only thing I
> >> can find is that /etc/mock/site-defaults.cfg has:
> >>
> >> # config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s
> >> -d %(home)s -n %(user)s' # Fedora/RedHat
> >>
> >> whereas /etc/mock/site-defaults.cfg.rpmnew has:
> >>
> >> # config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s
> >> -d %(home)s -N %(user)s' # Fedora/RedHat
> >
> > As an aside, the comment "Fedora/RedHat" is now untrue. It should
> > read "Fedora/RedHat > 6" perhaps.
> >
> >> So I suppose the default changed and is used for EL-5 builds.
> >>
> >> Given that my Fedora builders build EL-5 packages as well, how would I
> >> fix this? In Koji or in mock?
> >
> > You can override the defaults from site-defaults in your epel config
> > files. Adding:
> >
> > config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s -d %(home)s -n %(user)s'
> >
> > to epel-{4,5}*.cfg should make mock use the right command for EL
> > builds.
> >
>
> Koji writes out it's own mock configuration files though. As a result,
> /etc/mock/koji/*.cfg (per buildroot) uses the mock default -n/-N. Either
> one is incompatible with the other, and I have not found an alternative.
>
> No setting in any epel*-*.cfg in /etc/mock/. can change that wrt. to
> what Koji executes. Instead, I've removed the -n/-N in
> /etc/mock/site-defaults.cfg

I've submitted a new mock update that reverts the -n/-N change. Testing
+ karma would be appreciated.

https://admin.fedoraproject.org/updates/mock-1.0.7-1.el5

--
Jesse Keating
Fedora -- Freedom˛ is a feature!
identi.ca: http://identi.ca/jkeating
--
buildsys mailing list
buildsys@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/buildsys
 
Old 03-12-2010, 12:44 AM
Jeroen van Meeuwen
 
Default mock 1.0.5 failure

On 03/12/2010 12:55 AM, Jesse Keating wrote:
> On Sat, 2010-03-06 at 18:47 +0100, Jeroen van Meeuwen wrote:
>> On 03/06/2010 06:11 PM, Todd Zullinger wrote:
>>> Jeroen van Meeuwen wrote:
>>>> The builds I'm talking about are executed through Koji; The only thing I
>>>> can find is that /etc/mock/site-defaults.cfg has:
>>>>
>>>> # config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s
>>>> -d %(home)s -n %(user)s' # Fedora/RedHat
>>>>
>>>> whereas /etc/mock/site-defaults.cfg.rpmnew has:
>>>>
>>>> # config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s
>>>> -d %(home)s -N %(user)s' # Fedora/RedHat
>>>
>>> As an aside, the comment "Fedora/RedHat" is now untrue. It should
>>> read "Fedora/RedHat > 6" perhaps.
>>>
>>>> So I suppose the default changed and is used for EL-5 builds.
>>>>
>>>> Given that my Fedora builders build EL-5 packages as well, how would I
>>>> fix this? In Koji or in mock?
>>>
>>> You can override the defaults from site-defaults in your epel config
>>> files. Adding:
>>>
>>> config_opts['useradd'] = '/usr/sbin/useradd -m -u %(uid)s -g %(gid)s -d %(home)s -n %(user)s'
>>>
>>> to epel-{4,5}*.cfg should make mock use the right command for EL
>>> builds.
>>>
>>
>> Koji writes out it's own mock configuration files though. As a result,
>> /etc/mock/koji/*.cfg (per buildroot) uses the mock default -n/-N. Either
>> one is incompatible with the other, and I have not found an alternative.
>>
>> No setting in any epel*-*.cfg in /etc/mock/. can change that wrt. to
>> what Koji executes. Instead, I've removed the -n/-N in
>> /etc/mock/site-defaults.cfg
>
> I've submitted a new mock update that reverts the -n/-N change. Testing
> + karma would be appreciated.
>

If it's really reverted wouldn't that introduce the incompatibility
question with F-12+? Did you do something else then revert the -n/-N
change like remove -n/-N altogether maybe?

-- Jeroen
--
buildsys mailing list
buildsys@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/buildsys
 
Old 03-12-2010, 02:25 PM
Jesse Keating
 
Default mock 1.0.5 failure

On Fri, 2010-03-12 at 02:44 +0100, Jeroen van Meeuwen wrote:
> If it's really reverted wouldn't that introduce the incompatibility
> question with F-12+? Did you do something else then revert the -n/-N
> change like remove -n/-N altogether maybe?

Turns out I misread the original bug report. While -n is deprecated on
newer Fedoras, it hasn't stopped functioning yet, so the best solution
for now is to just use -n everywhere. If a time comes that -n doesn't
work on Fedora we'll have to revisit this, although I think I'd petition
upstream to never make that happen.

--
Jesse Keating
Fedora -- Freedom˛ is a feature!
identi.ca: http://identi.ca/jkeating
--
buildsys mailing list
buildsys@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/buildsys
 

Thread Tools




All times are GMT. The time now is 01:00 PM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright ©2007 - 2008, www.linux-archive.org