Re: [foreman-dev] Koji repo sync metadata problem

[Lukas Zapletal]

Sorry we had public holiday here.

We use mirrorselect with mrepo, sometimes it can pick slower mirror.
Would be safer to pick particular mirrors if we plan to sync
regularly, ideally something within AWS datacenters if there's any.

LZ

On Fri, Nov 17, 2017 at 4:53 PM, Eric D Helms  wrote:
> The update seems to have completed, with the new CentOS repository added to
> Koji and repositories regenerated. For the moment, the
> foreman-nightly*-rhel7-build tags have been updated to remove rhel as an
> external repository and have been replaced with CentOS (7.4). If you notice
> any oddities building, please let us know so we can work through them.
>
>
> Eric
>
> On Thu, Nov 16, 2017 at 7:14 PM, Eric D Helms  wrote:
>>
>> I started an mrepo run after adding CentOS in a screen session. The run
>> appears to be running a lot longer than previous and it's not clear to me
>> what exactly it's doing at present or whats safe to do. Lukas could you
>> check on it in your morning and see if you spot anything odd about why it
>> seems to have stalled or slowed down?
>>
>> On Thu, Nov 16, 2017 at 6:01 AM, Ewoud Kohl van Wijngaarden
>>  wrote:
>>>
>>> Those sound like good arguments, but if we don't test on older releases
>>> then we don't know for sure they work either. IMHO it's acceptable to
>>> explicitly only support EL 7.4 and up. We don't have the resources to
>>> support older versions especially since CentOS doesn't support that either.
>>> If users want verified working versions on EL < 7.4 then they should either
>>> help us properly support it or buy support from a vendor that does want to
>>> promise it works.
>>>
>>>
>>> On Thu, Nov 16, 2017 at 08:56:39AM +0100, Lukas Zapletal wrote:

 I would be against syncing regularly but I don't do much packaging so
 it's up to you. The problem I see is "random" breakage. You come to
 work on Monday after sync and if you don't realize the package was
 deleted from EPEL, you can burn hour or something to figure it out
 sometimes. This was pretty clear case but as you know there are
 situations (SCL) when it is not that clear.

 If we ever deploy new koji, I'd vote to avoid getting OS updates
 completely, just base channels and explicit updates when we need it.
 We need reproducible builds, with autosync we are not getting it.

 One note, if we sync CentOS or RHEL base to particular version,
 chances are that SELinux won't load on older releases. So basically we
 must be ready to do minimum requirements change here.

 Thanks for solving the problem.

 LZ

 On Wed, Nov 15, 2017 at 10:05 PM, Ewoud Kohl van Wijngaarden
  wrote:
>
> I think we should sync & update. CentOS has no concept of supported
> minor
> releases and we should be testing with a supported release.
>
>
> On Wed, Nov 15, 2017 at 03:57:40PM -0500, Eric D Helms wrote:
>>
>>
>> This now appears to be working again. One out standing question that
>> affects nodejs- packaging builds.
>>
>> As part of the RHEL 7.4 release, http-parser was removed from EPEL.
>> With
>> this latest round of Koji external repositories update we now have a
>> newer
>> EPEL with this package removed. We did not update our CentOS external
>> repository that would include this package. Not updating the base OS
>> has
>> been our policy for a while now it would seem. We have two options:
>>
>> 1) Sync and update CentOS
>> 2) Add http-parser to the overrides tag for foreman-nightly
>>
>> On Wed, Nov 15, 2017 at 2:11 PM, Eric D Helms 
>> wrote:
>>
>>> I had forgotten to run the repo-regen on Koji. That is finishing up
>>> now.
>>> I
>>> will run another test after and report back.
>>>
>>> On Wed, Nov 15, 2017 at 2:08 PM, Patrick Creech 
>>> wrote:
>>>
 On Wed, 2017-11-15 at 14:00 -0500, Patrick Creech wrote:
 > On Wed, 2017-11-15 at 19:56 +0100, Lukas Zapletal wrote:
 > > I do not have access here, can someone take a look? Are repodata
 > > present for EPEL7?
 >
 > If I have the correct path, it does not appear so:
 >
 > [root@koji repodata]# pwd
 > /mnt/koji/external-repos/epel7-x86_64/epel/repodata
 > [root@koji repodata]# ls
 > [root@koji repodata]#
 >

 I might not have, this path DOES have repodata:

 /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/repodata

 > > LZ
 > >
 > > On Wed, Nov 15, 2017 at 6:15 PM, Ewoud Kohl van Wijngaarden
 > >  wrote:
 > > > I still get errors when building:
 > > >
 > > > 

Re: [foreman-dev] Koji repo sync metadata problem

[Eric D Helms]

The update seems to have completed, with the new CentOS repository added to
Koji and repositories regenerated. For the moment, the
foreman-nightly*-rhel7-build tags have been updated to remove rhel as an
external repository and have been replaced with CentOS (7.4). If you notice
any oddities building, please let us know so we can work through them.


Eric

On Thu, Nov 16, 2017 at 7:14 PM, Eric D Helms  wrote:

> I started an mrepo run after adding CentOS in a screen session. The run
> appears to be running a lot longer than previous and it's not clear to me
> what exactly it's doing at present or whats safe to do. Lukas could you
> check on it in your morning and see if you spot anything odd about why it
> seems to have stalled or slowed down?
>
> On Thu, Nov 16, 2017 at 6:01 AM, Ewoud Kohl van Wijngaarden <
> ew...@kohlvanwijngaarden.nl> wrote:
>
>> Those sound like good arguments, but if we don't test on older releases
>> then we don't know for sure they work either. IMHO it's acceptable to
>> explicitly only support EL 7.4 and up. We don't have the resources to
>> support older versions especially since CentOS doesn't support that either.
>> If users want verified working versions on EL < 7.4 then they should either
>> help us properly support it or buy support from a vendor that does want to
>> promise it works.
>>
>>
>> On Thu, Nov 16, 2017 at 08:56:39AM +0100, Lukas Zapletal wrote:
>>
>>> I would be against syncing regularly but I don't do much packaging so
>>> it's up to you. The problem I see is "random" breakage. You come to
>>> work on Monday after sync and if you don't realize the package was
>>> deleted from EPEL, you can burn hour or something to figure it out
>>> sometimes. This was pretty clear case but as you know there are
>>> situations (SCL) when it is not that clear.
>>>
>>> If we ever deploy new koji, I'd vote to avoid getting OS updates
>>> completely, just base channels and explicit updates when we need it.
>>> We need reproducible builds, with autosync we are not getting it.
>>>
>>> One note, if we sync CentOS or RHEL base to particular version,
>>> chances are that SELinux won't load on older releases. So basically we
>>> must be ready to do minimum requirements change here.
>>>
>>> Thanks for solving the problem.
>>>
>>> LZ
>>>
>>> On Wed, Nov 15, 2017 at 10:05 PM, Ewoud Kohl van Wijngaarden
>>>  wrote:
>>>
 I think we should sync & update. CentOS has no concept of supported
 minor
 releases and we should be testing with a supported release.


 On Wed, Nov 15, 2017 at 03:57:40PM -0500, Eric D Helms wrote:

>
> This now appears to be working again. One out standing question that
> affects nodejs- packaging builds.
>
> As part of the RHEL 7.4 release, http-parser was removed from EPEL.
> With
> this latest round of Koji external repositories update we now have a
> newer
> EPEL with this package removed. We did not update our CentOS external
> repository that would include this package. Not updating the base OS
> has
> been our policy for a while now it would seem. We have two options:
>
> 1) Sync and update CentOS
> 2) Add http-parser to the overrides tag for foreman-nightly
>
> On Wed, Nov 15, 2017 at 2:11 PM, Eric D Helms 
> wrote:
>
> I had forgotten to run the repo-regen on Koji. That is finishing up
>> now.
>> I
>> will run another test after and report back.
>>
>> On Wed, Nov 15, 2017 at 2:08 PM, Patrick Creech 
>> wrote:
>>
>> On Wed, 2017-11-15 at 14:00 -0500, Patrick Creech wrote:
>>> > On Wed, 2017-11-15 at 19:56 +0100, Lukas Zapletal wrote:
>>> > > I do not have access here, can someone take a look? Are repodata
>>> > > present for EPEL7?
>>> >
>>> > If I have the correct path, it does not appear so:
>>> >
>>> > [root@koji repodata]# pwd
>>> > /mnt/koji/external-repos/epel7-x86_64/epel/repodata
>>> > [root@koji repodata]# ls
>>> > [root@koji repodata]#
>>> >
>>>
>>> I might not have, this path DOES have repodata:
>>>
>>> /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/repodata
>>>
>>> > > LZ
>>> > >
>>> > > On Wed, Nov 15, 2017 at 6:15 PM, Ewoud Kohl van Wijngaarden
>>> > >  wrote:
>>> > > > I still get errors when building:
>>> > > >
>>> > > > http://koji.katello.org/koji/taskinfo?taskID=52363
>>> > > > http://koji.katello.org/kojifiles/work/tasks/2363/52363/root
>>> .log
>>> > > >
>>> > > > DEBUG util.py:439:  Error downloading packages:
>>> > > > DEBUG util.py:439:1:nodejs-6.10.3-1.el7.x86_64: failed to
>>> retrieve
>>> > > > nodejs-6.10.3-1.el7.x86_64.rpm from build
>>> > > > DEBUG util.py:439:  error was [Errno 2] Local file does not
>>> exist:
>>> > > > 

Re: [foreman-dev] Koji repo sync metadata problem

[Eric D Helms]

I started an mrepo run after adding CentOS in a screen session. The run
appears to be running a lot longer than previous and it's not clear to me
what exactly it's doing at present or whats safe to do. Lukas could you
check on it in your morning and see if you spot anything odd about why it
seems to have stalled or slowed down?

On Thu, Nov 16, 2017 at 6:01 AM, Ewoud Kohl van Wijngaarden <
ew...@kohlvanwijngaarden.nl> wrote:

> Those sound like good arguments, but if we don't test on older releases
> then we don't know for sure they work either. IMHO it's acceptable to
> explicitly only support EL 7.4 and up. We don't have the resources to
> support older versions especially since CentOS doesn't support that either.
> If users want verified working versions on EL < 7.4 then they should either
> help us properly support it or buy support from a vendor that does want to
> promise it works.
>
>
> On Thu, Nov 16, 2017 at 08:56:39AM +0100, Lukas Zapletal wrote:
>
>> I would be against syncing regularly but I don't do much packaging so
>> it's up to you. The problem I see is "random" breakage. You come to
>> work on Monday after sync and if you don't realize the package was
>> deleted from EPEL, you can burn hour or something to figure it out
>> sometimes. This was pretty clear case but as you know there are
>> situations (SCL) when it is not that clear.
>>
>> If we ever deploy new koji, I'd vote to avoid getting OS updates
>> completely, just base channels and explicit updates when we need it.
>> We need reproducible builds, with autosync we are not getting it.
>>
>> One note, if we sync CentOS or RHEL base to particular version,
>> chances are that SELinux won't load on older releases. So basically we
>> must be ready to do minimum requirements change here.
>>
>> Thanks for solving the problem.
>>
>> LZ
>>
>> On Wed, Nov 15, 2017 at 10:05 PM, Ewoud Kohl van Wijngaarden
>>  wrote:
>>
>>> I think we should sync & update. CentOS has no concept of supported minor
>>> releases and we should be testing with a supported release.
>>>
>>>
>>> On Wed, Nov 15, 2017 at 03:57:40PM -0500, Eric D Helms wrote:
>>>

 This now appears to be working again. One out standing question that
 affects nodejs- packaging builds.

 As part of the RHEL 7.4 release, http-parser was removed from EPEL. With
 this latest round of Koji external repositories update we now have a
 newer
 EPEL with this package removed. We did not update our CentOS external
 repository that would include this package. Not updating the base OS has
 been our policy for a while now it would seem. We have two options:

 1) Sync and update CentOS
 2) Add http-parser to the overrides tag for foreman-nightly

 On Wed, Nov 15, 2017 at 2:11 PM, Eric D Helms 
 wrote:

 I had forgotten to run the repo-regen on Koji. That is finishing up now.
> I
> will run another test after and report back.
>
> On Wed, Nov 15, 2017 at 2:08 PM, Patrick Creech 
> wrote:
>
> On Wed, 2017-11-15 at 14:00 -0500, Patrick Creech wrote:
>> > On Wed, 2017-11-15 at 19:56 +0100, Lukas Zapletal wrote:
>> > > I do not have access here, can someone take a look? Are repodata
>> > > present for EPEL7?
>> >
>> > If I have the correct path, it does not appear so:
>> >
>> > [root@koji repodata]# pwd
>> > /mnt/koji/external-repos/epel7-x86_64/epel/repodata
>> > [root@koji repodata]# ls
>> > [root@koji repodata]#
>> >
>>
>> I might not have, this path DOES have repodata:
>>
>> /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/repodata
>>
>> > > LZ
>> > >
>> > > On Wed, Nov 15, 2017 at 6:15 PM, Ewoud Kohl van Wijngaarden
>> > >  wrote:
>> > > > I still get errors when building:
>> > > >
>> > > > http://koji.katello.org/koji/taskinfo?taskID=52363
>> > > > http://koji.katello.org/kojifiles/work/tasks/2363/52363/
>> root.log
>> > > >
>> > > > DEBUG util.py:439:  Error downloading packages:
>> > > > DEBUG util.py:439:1:nodejs-6.10.3-1.el7.x86_64: failed to
>> retrieve
>> > > > nodejs-6.10.3-1.el7.x86_64.rpm from build
>> > > > DEBUG util.py:439:  error was [Errno 2] Local file does not
>> exist:
>> > > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/nodejs-6
>> .10.3-1.el7.x86_64.rpm
>> > > > DEBUG util.py:439:http-parser-2.7.1-3.el7.x86_64: failed to
>> retrieve
>> > > > http-parser-2.7.1-3.el7.x86_64.rpm from build
>> > > > DEBUG util.py:439:  error was [Errno 2] Local file does not
>> exist:
>> > > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/http-par
>> ser-2.7.1-3.el7.x86_64.rpm
>> > > > DEBUG util.py:439:1:npm-3.10.10-1.6.10.3.1.el7.x86_64:
>> failed
>> to
>> > > > retrieve 

Re: [foreman-dev] Koji repo sync metadata problem

[Ewoud Kohl van Wijngaarden]

Those sound like good arguments, but if we don't test on older releases 
then we don't know for sure they work either. IMHO it's acceptable to 
explicitly only support EL 7.4 and up. We don't have the resources to 
support older versions especially since CentOS doesn't support that 
either. If users want verified working versions on EL < 7.4 then they 
should either help us properly support it or buy support from a vendor 
that does want to promise it works.


On Thu, Nov 16, 2017 at 08:56:39AM +0100, Lukas Zapletal wrote:

I would be against syncing regularly but I don't do much packaging so
it's up to you. The problem I see is "random" breakage. You come to
work on Monday after sync and if you don't realize the package was
deleted from EPEL, you can burn hour or something to figure it out
sometimes. This was pretty clear case but as you know there are
situations (SCL) when it is not that clear.

If we ever deploy new koji, I'd vote to avoid getting OS updates
completely, just base channels and explicit updates when we need it.
We need reproducible builds, with autosync we are not getting it.

One note, if we sync CentOS or RHEL base to particular version,
chances are that SELinux won't load on older releases. So basically we
must be ready to do minimum requirements change here.

Thanks for solving the problem.

LZ

On Wed, Nov 15, 2017 at 10:05 PM, Ewoud Kohl van Wijngaarden
 wrote:

I think we should sync & update. CentOS has no concept of supported minor
releases and we should be testing with a supported release.


On Wed, Nov 15, 2017 at 03:57:40PM -0500, Eric D Helms wrote:


This now appears to be working again. One out standing question that
affects nodejs- packaging builds.

As part of the RHEL 7.4 release, http-parser was removed from EPEL. With
this latest round of Koji external repositories update we now have a newer
EPEL with this package removed. We did not update our CentOS external
repository that would include this package. Not updating the base OS has
been our policy for a while now it would seem. We have two options:

1) Sync and update CentOS
2) Add http-parser to the overrides tag for foreman-nightly

On Wed, Nov 15, 2017 at 2:11 PM, Eric D Helms 
wrote:


I had forgotten to run the repo-regen on Koji. That is finishing up now.
I
will run another test after and report back.

On Wed, Nov 15, 2017 at 2:08 PM, Patrick Creech 
wrote:


On Wed, 2017-11-15 at 14:00 -0500, Patrick Creech wrote:
> On Wed, 2017-11-15 at 19:56 +0100, Lukas Zapletal wrote:
> > I do not have access here, can someone take a look? Are repodata
> > present for EPEL7?
>
> If I have the correct path, it does not appear so:
>
> [root@koji repodata]# pwd
> /mnt/koji/external-repos/epel7-x86_64/epel/repodata
> [root@koji repodata]# ls
> [root@koji repodata]#
>

I might not have, this path DOES have repodata:

/mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/repodata

> > LZ
> >
> > On Wed, Nov 15, 2017 at 6:15 PM, Ewoud Kohl van Wijngaarden
> >  wrote:
> > > I still get errors when building:
> > >
> > > http://koji.katello.org/koji/taskinfo?taskID=52363
> > > http://koji.katello.org/kojifiles/work/tasks/2363/52363/root.log
> > >
> > > DEBUG util.py:439:  Error downloading packages:
> > > DEBUG util.py:439:1:nodejs-6.10.3-1.el7.x86_64: failed to
retrieve
> > > nodejs-6.10.3-1.el7.x86_64.rpm from build
> > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
> > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/nodejs-6
.10.3-1.el7.x86_64.rpm
> > > DEBUG util.py:439:http-parser-2.7.1-3.el7.x86_64: failed to
retrieve
> > > http-parser-2.7.1-3.el7.x86_64.rpm from build
> > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
> > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/http-par
ser-2.7.1-3.el7.x86_64.rpm
> > > DEBUG util.py:439:1:npm-3.10.10-1.6.10.3.1.el7.x86_64: failed
to
> > > retrieve npm-3.10.10-1.6.10.3.1.el7.x86_64.rpm from build
> > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
> > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/npm-3.10
.10-1.6.10.3.1.el7.x86_64.rpm
> > >
> > >
> > > On Wed, Nov 15, 2017 at 11:44:24AM -0500, Eric D Helms wrote:
> > > >
> > > > Looks like everything is back up and working as expected. For
packagers,
> > > > keep in mind that this updated some of our external repositories
to their
> > > > latest versions if you see any oddities. This also means that
Fedora 27 is
> > > > available as an external repository for Pulp and Katello
> > > > clients.
> > > >
> > > > On Wed, Nov 15, 2017 at 9:32 AM, Lukas Zapletal
> > > > 
wrote:
> > > >
> > > > > Hey,
> > > > >
> > > > > when I initiated mrepo -n (dry run) this morning to see how
mrepo
> > > > > works on our koji in order to test if we are able to add
> > > > > Fedora
27 for
> > > > > Pulp, I learned that this "dry run" is actually a 

Re: [foreman-dev] Koji repo sync metadata problem

[Lukas Zapletal]

I would be against syncing regularly but I don't do much packaging so
it's up to you. The problem I see is "random" breakage. You come to
work on Monday after sync and if you don't realize the package was
deleted from EPEL, you can burn hour or something to figure it out
sometimes. This was pretty clear case but as you know there are
situations (SCL) when it is not that clear.

If we ever deploy new koji, I'd vote to avoid getting OS updates
completely, just base channels and explicit updates when we need it.
We need reproducible builds, with autosync we are not getting it.

One note, if we sync CentOS or RHEL base to particular version,
chances are that SELinux won't load on older releases. So basically we
must be ready to do minimum requirements change here.

Thanks for solving the problem.

LZ

On Wed, Nov 15, 2017 at 10:05 PM, Ewoud Kohl van Wijngaarden
 wrote:
> I think we should sync & update. CentOS has no concept of supported minor
> releases and we should be testing with a supported release.
>
>
> On Wed, Nov 15, 2017 at 03:57:40PM -0500, Eric D Helms wrote:
>>
>> This now appears to be working again. One out standing question that
>> affects nodejs- packaging builds.
>>
>> As part of the RHEL 7.4 release, http-parser was removed from EPEL. With
>> this latest round of Koji external repositories update we now have a newer
>> EPEL with this package removed. We did not update our CentOS external
>> repository that would include this package. Not updating the base OS has
>> been our policy for a while now it would seem. We have two options:
>>
>> 1) Sync and update CentOS
>> 2) Add http-parser to the overrides tag for foreman-nightly
>>
>> On Wed, Nov 15, 2017 at 2:11 PM, Eric D Helms 
>> wrote:
>>
>>> I had forgotten to run the repo-regen on Koji. That is finishing up now.
>>> I
>>> will run another test after and report back.
>>>
>>> On Wed, Nov 15, 2017 at 2:08 PM, Patrick Creech 
>>> wrote:
>>>
 On Wed, 2017-11-15 at 14:00 -0500, Patrick Creech wrote:
 > On Wed, 2017-11-15 at 19:56 +0100, Lukas Zapletal wrote:
 > > I do not have access here, can someone take a look? Are repodata
 > > present for EPEL7?
 >
 > If I have the correct path, it does not appear so:
 >
 > [root@koji repodata]# pwd
 > /mnt/koji/external-repos/epel7-x86_64/epel/repodata
 > [root@koji repodata]# ls
 > [root@koji repodata]#
 >

 I might not have, this path DOES have repodata:

 /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/repodata

 > > LZ
 > >
 > > On Wed, Nov 15, 2017 at 6:15 PM, Ewoud Kohl van Wijngaarden
 > >  wrote:
 > > > I still get errors when building:
 > > >
 > > > http://koji.katello.org/koji/taskinfo?taskID=52363
 > > > http://koji.katello.org/kojifiles/work/tasks/2363/52363/root.log
 > > >
 > > > DEBUG util.py:439:  Error downloading packages:
 > > > DEBUG util.py:439:1:nodejs-6.10.3-1.el7.x86_64: failed to
 retrieve
 > > > nodejs-6.10.3-1.el7.x86_64.rpm from build
 > > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
 > > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/nodejs-6
 .10.3-1.el7.x86_64.rpm
 > > > DEBUG util.py:439:http-parser-2.7.1-3.el7.x86_64: failed to
 retrieve
 > > > http-parser-2.7.1-3.el7.x86_64.rpm from build
 > > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
 > > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/http-par
 ser-2.7.1-3.el7.x86_64.rpm
 > > > DEBUG util.py:439:1:npm-3.10.10-1.6.10.3.1.el7.x86_64: failed
 to
 > > > retrieve npm-3.10.10-1.6.10.3.1.el7.x86_64.rpm from build
 > > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
 > > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/npm-3.10
 .10-1.6.10.3.1.el7.x86_64.rpm
 > > >
 > > >
 > > > On Wed, Nov 15, 2017 at 11:44:24AM -0500, Eric D Helms wrote:
 > > > >
 > > > > Looks like everything is back up and working as expected. For
 packagers,
 > > > > keep in mind that this updated some of our external repositories
 to their
 > > > > latest versions if you see any oddities. This also means that
 Fedora 27 is
 > > > > available as an external repository for Pulp and Katello
 > > > > clients.
 > > > >
 > > > > On Wed, Nov 15, 2017 at 9:32 AM, Lukas Zapletal
 > > > > 
 wrote:
 > > > >
 > > > > > Hey,
 > > > > >
 > > > > > when I initiated mrepo -n (dry run) this morning to see how
 mrepo
 > > > > > works on our koji in order to test if we are able to add
 > > > > > Fedora
 27 for
 > > > > > Pulp, I learned that this "dry run" is actually a real run and
 mrepo
 > > > > > initiated full resync of our content without metadata
 regeneration.

Re: [foreman-dev] Koji repo sync metadata problem

[Ewoud Kohl van Wijngaarden]

I think we should sync & update. CentOS has no concept of supported 
minor releases and we should be testing with a supported release.


On Wed, Nov 15, 2017 at 03:57:40PM -0500, Eric D Helms wrote:

This now appears to be working again. One out standing question that
affects nodejs- packaging builds.

As part of the RHEL 7.4 release, http-parser was removed from EPEL. With
this latest round of Koji external repositories update we now have a newer
EPEL with this package removed. We did not update our CentOS external
repository that would include this package. Not updating the base OS has
been our policy for a while now it would seem. We have two options:

1) Sync and update CentOS
2) Add http-parser to the overrides tag for foreman-nightly

On Wed, Nov 15, 2017 at 2:11 PM, Eric D Helms  wrote:


I had forgotten to run the repo-regen on Koji. That is finishing up now. I
will run another test after and report back.

On Wed, Nov 15, 2017 at 2:08 PM, Patrick Creech 
wrote:


On Wed, 2017-11-15 at 14:00 -0500, Patrick Creech wrote:
> On Wed, 2017-11-15 at 19:56 +0100, Lukas Zapletal wrote:
> > I do not have access here, can someone take a look? Are repodata
> > present for EPEL7?
>
> If I have the correct path, it does not appear so:
>
> [root@koji repodata]# pwd
> /mnt/koji/external-repos/epel7-x86_64/epel/repodata
> [root@koji repodata]# ls
> [root@koji repodata]#
>

I might not have, this path DOES have repodata:

/mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/repodata

> > LZ
> >
> > On Wed, Nov 15, 2017 at 6:15 PM, Ewoud Kohl van Wijngaarden
> >  wrote:
> > > I still get errors when building:
> > >
> > > http://koji.katello.org/koji/taskinfo?taskID=52363
> > > http://koji.katello.org/kojifiles/work/tasks/2363/52363/root.log
> > >
> > > DEBUG util.py:439:  Error downloading packages:
> > > DEBUG util.py:439:1:nodejs-6.10.3-1.el7.x86_64: failed to
retrieve
> > > nodejs-6.10.3-1.el7.x86_64.rpm from build
> > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
> > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/nodejs-6
.10.3-1.el7.x86_64.rpm
> > > DEBUG util.py:439:http-parser-2.7.1-3.el7.x86_64: failed to
retrieve
> > > http-parser-2.7.1-3.el7.x86_64.rpm from build
> > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
> > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/http-par
ser-2.7.1-3.el7.x86_64.rpm
> > > DEBUG util.py:439:1:npm-3.10.10-1.6.10.3.1.el7.x86_64: failed
to
> > > retrieve npm-3.10.10-1.6.10.3.1.el7.x86_64.rpm from build
> > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
> > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/npm-3.10
.10-1.6.10.3.1.el7.x86_64.rpm
> > >
> > >
> > > On Wed, Nov 15, 2017 at 11:44:24AM -0500, Eric D Helms wrote:
> > > >
> > > > Looks like everything is back up and working as expected. For
packagers,
> > > > keep in mind that this updated some of our external repositories
to their
> > > > latest versions if you see any oddities. This also means that
Fedora 27 is
> > > > available as an external repository for Pulp and Katello clients.
> > > >
> > > > On Wed, Nov 15, 2017 at 9:32 AM, Lukas Zapletal 
wrote:
> > > >
> > > > > Hey,
> > > > >
> > > > > when I initiated mrepo -n (dry run) this morning to see how
mrepo
> > > > > works on our koji in order to test if we are able to add Fedora
27 for
> > > > > Pulp, I learned that this "dry run" is actually a real run and
mrepo
> > > > > initiated full resync of our content without metadata
regeneration.
> > > > >
> > > > > This rendered our external repositories unusable as all
metadata were
> > > > > deleted. All builds will fail today with missing dependencies.
> > > > >
> > > > > The plan: we need to let mrepo finish "dry run" downloading of
Fedora
> > > > > 27 and then we need to reexecute it with proper flags:
> > > > >
> > > > > cat /usr/local/bin/koji-sync-external-repos
> > > > > #!/bin/bash
> > > > > mrepo -qugf
> > > > > koji list-targets --quiet | awk '{print $2}' | sort -u | xargs
-I '{}'
> > > > > koji regen-repo --nowait '{}'
> > > > >
> > > > > Note the -u -g -f flags for mrepo, this will cause all metadata
to be
> > > > > regenerated. Once this command is done, koji will see missing
> > > > > dependencies to appear.
> > > > >
> > > > > The dry run of mrepo is running in screen and it is about to
finish
> > > > > hopefully soon. I will defer this to Eric in case it won't
finish by
> > > > > EMEA EOB.
> > > > >
> > > > > Sorry for inconvenience, but dry run should be dry run, right.
> > > > >
> > > > > --
> > > > > Later,
> > > > >   Lukas @lzap Zapletal
> > > > >
> > > > > --
> > > > > You received this message because you are subscribed to the
Google Groups
> > > > > "foreman-dev" group.
> > > > > To unsubscribe from this group and stop receiving emails from
it, send an
> > > > > email to foreman-dev+unsubscr...@googlegroups.com.

Re: [foreman-dev] Koji repo sync metadata problem

[Eric D Helms]

This now appears to be working again. One out standing question that
affects nodejs- packaging builds.

As part of the RHEL 7.4 release, http-parser was removed from EPEL. With
this latest round of Koji external repositories update we now have a newer
EPEL with this package removed. We did not update our CentOS external
repository that would include this package. Not updating the base OS has
been our policy for a while now it would seem. We have two options:

 1) Sync and update CentOS
 2) Add http-parser to the overrides tag for foreman-nightly

On Wed, Nov 15, 2017 at 2:11 PM, Eric D Helms  wrote:

> I had forgotten to run the repo-regen on Koji. That is finishing up now. I
> will run another test after and report back.
>
> On Wed, Nov 15, 2017 at 2:08 PM, Patrick Creech 
> wrote:
>
>> On Wed, 2017-11-15 at 14:00 -0500, Patrick Creech wrote:
>> > On Wed, 2017-11-15 at 19:56 +0100, Lukas Zapletal wrote:
>> > > I do not have access here, can someone take a look? Are repodata
>> > > present for EPEL7?
>> >
>> > If I have the correct path, it does not appear so:
>> >
>> > [root@koji repodata]# pwd
>> > /mnt/koji/external-repos/epel7-x86_64/epel/repodata
>> > [root@koji repodata]# ls
>> > [root@koji repodata]#
>> >
>>
>> I might not have, this path DOES have repodata:
>>
>> /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/repodata
>>
>> > > LZ
>> > >
>> > > On Wed, Nov 15, 2017 at 6:15 PM, Ewoud Kohl van Wijngaarden
>> > >  wrote:
>> > > > I still get errors when building:
>> > > >
>> > > > http://koji.katello.org/koji/taskinfo?taskID=52363
>> > > > http://koji.katello.org/kojifiles/work/tasks/2363/52363/root.log
>> > > >
>> > > > DEBUG util.py:439:  Error downloading packages:
>> > > > DEBUG util.py:439:1:nodejs-6.10.3-1.el7.x86_64: failed to
>> retrieve
>> > > > nodejs-6.10.3-1.el7.x86_64.rpm from build
>> > > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
>> > > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/nodejs-6
>> .10.3-1.el7.x86_64.rpm
>> > > > DEBUG util.py:439:http-parser-2.7.1-3.el7.x86_64: failed to
>> retrieve
>> > > > http-parser-2.7.1-3.el7.x86_64.rpm from build
>> > > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
>> > > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/http-par
>> ser-2.7.1-3.el7.x86_64.rpm
>> > > > DEBUG util.py:439:1:npm-3.10.10-1.6.10.3.1.el7.x86_64: failed
>> to
>> > > > retrieve npm-3.10.10-1.6.10.3.1.el7.x86_64.rpm from build
>> > > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
>> > > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/npm-3.10
>> .10-1.6.10.3.1.el7.x86_64.rpm
>> > > >
>> > > >
>> > > > On Wed, Nov 15, 2017 at 11:44:24AM -0500, Eric D Helms wrote:
>> > > > >
>> > > > > Looks like everything is back up and working as expected. For
>> packagers,
>> > > > > keep in mind that this updated some of our external repositories
>> to their
>> > > > > latest versions if you see any oddities. This also means that
>> Fedora 27 is
>> > > > > available as an external repository for Pulp and Katello clients.
>> > > > >
>> > > > > On Wed, Nov 15, 2017 at 9:32 AM, Lukas Zapletal 
>> wrote:
>> > > > >
>> > > > > > Hey,
>> > > > > >
>> > > > > > when I initiated mrepo -n (dry run) this morning to see how
>> mrepo
>> > > > > > works on our koji in order to test if we are able to add Fedora
>> 27 for
>> > > > > > Pulp, I learned that this "dry run" is actually a real run and
>> mrepo
>> > > > > > initiated full resync of our content without metadata
>> regeneration.
>> > > > > >
>> > > > > > This rendered our external repositories unusable as all
>> metadata were
>> > > > > > deleted. All builds will fail today with missing dependencies.
>> > > > > >
>> > > > > > The plan: we need to let mrepo finish "dry run" downloading of
>> Fedora
>> > > > > > 27 and then we need to reexecute it with proper flags:
>> > > > > >
>> > > > > > cat /usr/local/bin/koji-sync-external-repos
>> > > > > > #!/bin/bash
>> > > > > > mrepo -qugf
>> > > > > > koji list-targets --quiet | awk '{print $2}' | sort -u | xargs
>> -I '{}'
>> > > > > > koji regen-repo --nowait '{}'
>> > > > > >
>> > > > > > Note the -u -g -f flags for mrepo, this will cause all metadata
>> to be
>> > > > > > regenerated. Once this command is done, koji will see missing
>> > > > > > dependencies to appear.
>> > > > > >
>> > > > > > The dry run of mrepo is running in screen and it is about to
>> finish
>> > > > > > hopefully soon. I will defer this to Eric in case it won't
>> finish by
>> > > > > > EMEA EOB.
>> > > > > >
>> > > > > > Sorry for inconvenience, but dry run should be dry run, right.
>> > > > > >
>> > > > > > --
>> > > > > > Later,
>> > > > > >   Lukas @lzap Zapletal
>> > > > > >
>> > > > > > --
>> > > > > > You received this message because you are subscribed to the
>> Google Groups
>> > > > > > "foreman-dev" group.
>> > > > > > To 

Re: [foreman-dev] Koji repo sync metadata problem

[Eric D Helms]

I had forgotten to run the repo-regen on Koji. That is finishing up now. I
will run another test after and report back.

On Wed, Nov 15, 2017 at 2:08 PM, Patrick Creech  wrote:

> On Wed, 2017-11-15 at 14:00 -0500, Patrick Creech wrote:
> > On Wed, 2017-11-15 at 19:56 +0100, Lukas Zapletal wrote:
> > > I do not have access here, can someone take a look? Are repodata
> > > present for EPEL7?
> >
> > If I have the correct path, it does not appear so:
> >
> > [root@koji repodata]# pwd
> > /mnt/koji/external-repos/epel7-x86_64/epel/repodata
> > [root@koji repodata]# ls
> > [root@koji repodata]#
> >
>
> I might not have, this path DOES have repodata:
>
> /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/repodata
>
> > > LZ
> > >
> > > On Wed, Nov 15, 2017 at 6:15 PM, Ewoud Kohl van Wijngaarden
> > >  wrote:
> > > > I still get errors when building:
> > > >
> > > > http://koji.katello.org/koji/taskinfo?taskID=52363
> > > > http://koji.katello.org/kojifiles/work/tasks/2363/52363/root.log
> > > >
> > > > DEBUG util.py:439:  Error downloading packages:
> > > > DEBUG util.py:439:1:nodejs-6.10.3-1.el7.x86_64: failed to
> retrieve
> > > > nodejs-6.10.3-1.el7.x86_64.rpm from build
> > > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
> > > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/nodejs-
> 6.10.3-1.el7.x86_64.rpm
> > > > DEBUG util.py:439:http-parser-2.7.1-3.el7.x86_64: failed to
> retrieve
> > > > http-parser-2.7.1-3.el7.x86_64.rpm from build
> > > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
> > > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/http-
> parser-2.7.1-3.el7.x86_64.rpm
> > > > DEBUG util.py:439:1:npm-3.10.10-1.6.10.3.1.el7.x86_64: failed to
> > > > retrieve npm-3.10.10-1.6.10.3.1.el7.x86_64.rpm from build
> > > > DEBUG util.py:439:  error was [Errno 2] Local file does not exist:
> > > > /mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/npm-3.
> 10.10-1.6.10.3.1.el7.x86_64.rpm
> > > >
> > > >
> > > > On Wed, Nov 15, 2017 at 11:44:24AM -0500, Eric D Helms wrote:
> > > > >
> > > > > Looks like everything is back up and working as expected. For
> packagers,
> > > > > keep in mind that this updated some of our external repositories
> to their
> > > > > latest versions if you see any oddities. This also means that
> Fedora 27 is
> > > > > available as an external repository for Pulp and Katello clients.
> > > > >
> > > > > On Wed, Nov 15, 2017 at 9:32 AM, Lukas Zapletal 
> wrote:
> > > > >
> > > > > > Hey,
> > > > > >
> > > > > > when I initiated mrepo -n (dry run) this morning to see how mrepo
> > > > > > works on our koji in order to test if we are able to add Fedora
> 27 for
> > > > > > Pulp, I learned that this "dry run" is actually a real run and
> mrepo
> > > > > > initiated full resync of our content without metadata
> regeneration.
> > > > > >
> > > > > > This rendered our external repositories unusable as all metadata
> were
> > > > > > deleted. All builds will fail today with missing dependencies.
> > > > > >
> > > > > > The plan: we need to let mrepo finish "dry run" downloading of
> Fedora
> > > > > > 27 and then we need to reexecute it with proper flags:
> > > > > >
> > > > > > cat /usr/local/bin/koji-sync-external-repos
> > > > > > #!/bin/bash
> > > > > > mrepo -qugf
> > > > > > koji list-targets --quiet | awk '{print $2}' | sort -u | xargs
> -I '{}'
> > > > > > koji regen-repo --nowait '{}'
> > > > > >
> > > > > > Note the -u -g -f flags for mrepo, this will cause all metadata
> to be
> > > > > > regenerated. Once this command is done, koji will see missing
> > > > > > dependencies to appear.
> > > > > >
> > > > > > The dry run of mrepo is running in screen and it is about to
> finish
> > > > > > hopefully soon. I will defer this to Eric in case it won't
> finish by
> > > > > > EMEA EOB.
> > > > > >
> > > > > > Sorry for inconvenience, but dry run should be dry run, right.
> > > > > >
> > > > > > --
> > > > > > Later,
> > > > > >   Lukas @lzap Zapletal
> > > > > >
> > > > > > --
> > > > > > You received this message because you are subscribed to the
> Google Groups
> > > > > > "foreman-dev" group.
> > > > > > To unsubscribe from this group and stop receiving emails from
> it, send an
> > > > > > email to foreman-dev+unsubscr...@googlegroups.com.
> > > > > > For more options, visit https://groups.google.com/d/optout.
> > > > > >
> > > > >
> > > > >
> > > > >
> > > > > --
> > > > > Eric D. Helms
> > > > > Red Hat Engineering
> > > > >
> > > > > --
> > > > > You received this message because you are subscribed to the Google
> Groups
> > > > > "foreman-dev" group.
> > > > > To unsubscribe from this group and stop receiving emails from it,
> send an
> > > > > email to foreman-dev+unsubscr...@googlegroups.com.
> > > > > For more options, visit https://groups.google.com/d/optout.
> > > >
> > > >
> > > > --
> > > > You received this message because you are 

Re: [foreman-dev] Koji repo sync metadata problem

[Ewoud Kohl van Wijngaarden]

I still get errors when building:

http://koji.katello.org/koji/taskinfo?taskID=52363
http://koji.katello.org/kojifiles/work/tasks/2363/52363/root.log

DEBUG util.py:439:  Error downloading packages:
DEBUG util.py:439:1:nodejs-6.10.3-1.el7.x86_64: failed to retrieve 
nodejs-6.10.3-1.el7.x86_64.rpm from build
DEBUG util.py:439:  error was [Errno 2] Local file does not exist: 
/mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/nodejs-6.10.3-1.el7.x86_64.rpm
DEBUG util.py:439:http-parser-2.7.1-3.el7.x86_64: failed to retrieve 
http-parser-2.7.1-3.el7.x86_64.rpm from build
DEBUG util.py:439:  error was [Errno 2] Local file does not exist: 
/mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/http-parser-2.7.1-3.el7.x86_64.rpm
DEBUG util.py:439:1:npm-3.10.10-1.6.10.3.1.el7.x86_64: failed to retrieve 
npm-3.10.10-1.6.10.3.1.el7.x86_64.rpm from build
DEBUG util.py:439:  error was [Errno 2] Local file does not exist: 
/mnt/koji/external-repos/www/epel7-x86_64/RPMS.epel/npm-3.10.10-1.6.10.3.1.el7.x86_64.rpm

On Wed, Nov 15, 2017 at 11:44:24AM -0500, Eric D Helms wrote:

Looks like everything is back up and working as expected. For packagers,
keep in mind that this updated some of our external repositories to their
latest versions if you see any oddities. This also means that Fedora 27 is
available as an external repository for Pulp and Katello clients.

On Wed, Nov 15, 2017 at 9:32 AM, Lukas Zapletal  wrote:


Hey,

when I initiated mrepo -n (dry run) this morning to see how mrepo
works on our koji in order to test if we are able to add Fedora 27 for
Pulp, I learned that this "dry run" is actually a real run and mrepo
initiated full resync of our content without metadata regeneration.

This rendered our external repositories unusable as all metadata were
deleted. All builds will fail today with missing dependencies.

The plan: we need to let mrepo finish "dry run" downloading of Fedora
27 and then we need to reexecute it with proper flags:

cat /usr/local/bin/koji-sync-external-repos
#!/bin/bash
mrepo -qugf
koji list-targets --quiet | awk '{print $2}' | sort -u | xargs -I '{}'
koji regen-repo --nowait '{}'

Note the -u -g -f flags for mrepo, this will cause all metadata to be
regenerated. Once this command is done, koji will see missing
dependencies to appear.

The dry run of mrepo is running in screen and it is about to finish
hopefully soon. I will defer this to Eric in case it won't finish by
EMEA EOB.

Sorry for inconvenience, but dry run should be dry run, right.

--
Later,
  Lukas @lzap Zapletal

--
You received this message because you are subscribed to the Google Groups
"foreman-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an
email to foreman-dev+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.





--
Eric D. Helms
Red Hat Engineering

--
You received this message because you are subscribed to the Google Groups 
"foreman-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to foreman-dev+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.


--
You received this message because you are subscribed to the Google Groups 
"foreman-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to foreman-dev+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Re: [foreman-dev] Koji repo sync metadata problem

[Eric D Helms]

Looks like everything is back up and working as expected. For packagers,
keep in mind that this updated some of our external repositories to their
latest versions if you see any oddities. This also means that Fedora 27 is
available as an external repository for Pulp and Katello clients.

On Wed, Nov 15, 2017 at 9:32 AM, Lukas Zapletal  wrote:

> Hey,
>
> when I initiated mrepo -n (dry run) this morning to see how mrepo
> works on our koji in order to test if we are able to add Fedora 27 for
> Pulp, I learned that this "dry run" is actually a real run and mrepo
> initiated full resync of our content without metadata regeneration.
>
> This rendered our external repositories unusable as all metadata were
> deleted. All builds will fail today with missing dependencies.
>
> The plan: we need to let mrepo finish "dry run" downloading of Fedora
> 27 and then we need to reexecute it with proper flags:
>
> cat /usr/local/bin/koji-sync-external-repos
> #!/bin/bash
> mrepo -qugf
> koji list-targets --quiet | awk '{print $2}' | sort -u | xargs -I '{}'
> koji regen-repo --nowait '{}'
>
> Note the -u -g -f flags for mrepo, this will cause all metadata to be
> regenerated. Once this command is done, koji will see missing
> dependencies to appear.
>
> The dry run of mrepo is running in screen and it is about to finish
> hopefully soon. I will defer this to Eric in case it won't finish by
> EMEA EOB.
>
> Sorry for inconvenience, but dry run should be dry run, right.
>
> --
> Later,
>   Lukas @lzap Zapletal
>
> --
> You received this message because you are subscribed to the Google Groups
> "foreman-dev" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to foreman-dev+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>



-- 
Eric D. Helms
Red Hat Engineering

-- 
You received this message because you are subscribed to the Google Groups 
"foreman-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to foreman-dev+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[foreman-dev] Koji repo sync metadata problem

[Lukas Zapletal]

Hey,

when I initiated mrepo -n (dry run) this morning to see how mrepo
works on our koji in order to test if we are able to add Fedora 27 for
Pulp, I learned that this "dry run" is actually a real run and mrepo
initiated full resync of our content without metadata regeneration.

This rendered our external repositories unusable as all metadata were
deleted. All builds will fail today with missing dependencies.

The plan: we need to let mrepo finish "dry run" downloading of Fedora
27 and then we need to reexecute it with proper flags:

cat /usr/local/bin/koji-sync-external-repos
#!/bin/bash
mrepo -qugf
koji list-targets --quiet | awk '{print $2}' | sort -u | xargs -I '{}'
koji regen-repo --nowait '{}'

Note the -u -g -f flags for mrepo, this will cause all metadata to be
regenerated. Once this command is done, koji will see missing
dependencies to appear.

The dry run of mrepo is running in screen and it is about to finish
hopefully soon. I will defer this to Eric in case it won't finish by
EMEA EOB.

Sorry for inconvenience, but dry run should be dry run, right.

-- 
Later,
  Lukas @lzap Zapletal

-- 
You received this message because you are subscribed to the Google Groups 
"foreman-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to foreman-dev+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.