Koji / epel-7 - What's going on?

[Frantisek Zatloukal]

Hi,
I am currently trying to build new resultsdb[0] into epel7-candidate.
However, I am facing some (weird?) behavior around python-flask,

I am able to do a successful build of resultsdb (depending on python-flask)
in mock epel-7 and koji scratch epel-7 [1]. However, it is failing on
not-available python-flask in koji to epel-7-candidate [2].

Am I overlooking something?

Thanks very much!

[0] https://src.fedoraproject.org/rpms/resultsdb/blob/epel7/f/resultsdb.spec
[1] https://koji.fedoraproject.org/koji/taskinfo?taskID=31025152
[2] https://koji.fedoraproject.org/koji/taskinfo?taskID=31025187

[Kevin Fenzi]

On 11/20/18 2:39 PM, Frantisek Zatloukal wrote:
> Hi,
> I am currently trying to build new resultsdb[0] into epel7-candidate.
> However, I am facing some (weird?) behavior around python-flask,
>
> I am able to do a successful build of resultsdb (depending on python-flask)
> in mock epel-7 and koji scratch epel-7 [1]. However, it is failing on
> not-available python-flask in koji to epel-7-candidate [2].
>
> Am I overlooking something?
>
> Thanks very much!
>
> [0] https://src.fedoraproject.org/rpms/resultsdb/blob/epel7/f/resultsdb.spec
> [1] https://koji.fedoraproject.org/koji/taskinfo?taskID=31025152
> [2] https://koji.fedoraproject.org/koji/taskinfo?taskID=31025187

flask was pulled into rhel-extras (for docker distribution I think) and
when it was, it was only published for x86_64.

Your package is noarch, but flask is not.

So, you happened to get a x86_64 builder with your scratch build, but
your real build got a ppc builder where flask is not available. ;(

I think you should be able to pass "Exclusivearch: noarch x86_64" and
have it work. If not, you can just resubmit your build until you get a
x86_64 builder.

kevin

[Frantisek Zatloukal]

On Tue, Nov 20, 2018 at 11:52 PM Kevin Fenzi  wrote:
> On 11/20/18 2:39 PM, Frantisek Zatloukal wrote:
> > Hi,
> > I am currently trying to build new resultsdb[0] into epel7-candidate.
> > However, I am facing some (weird?) behavior around python-flask,
> >
> > I am able to do a successful build of resultsdb (depending on
>
> python-flask)
> > in mock epel-7 and koji scratch epel-7 [1]. However, it is failing on
> > not-available python-flask in koji to epel-7-candidate [2].
> >
> > Am I overlooking something?
> >
> > Thanks very much!
> >
> > [0]
>
> https://src.fedoraproject.org/rpms/resultsdb/blob/epel7/f/resultsdb.spec
> > [1] https://koji.fedoraproject.org/koji/taskinfo?taskID=31025152
> > [2] https://koji.fedoraproject.org/koji/taskinfo?taskID=31025187
>
> flask was pulled into rhel-extras (for docker distribution I think) and
> when it was, it was only published for x86_64.
>
> Your package is noarch, but flask is not.
>
> So, you happened to get a x86_64 builder with your scratch build, but
> your real build got a ppc builder where flask is not available. ;(
>
> I think you should be able to pass "Exclusivearch: noarch x86_64" and
> have it work. If not, you can just resubmit your build until you get a
> x86_64 builder.
>
> kevin
>
> infrastructure mailing list -- [e-mail]
> To unsubscribe send an email to
> [e-mail]
> Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives:
> https://lists.fedoraproject.org/archives/list/[e-mail]...

I'll try it, thanks for the quick reply!

[Vít Ondruch]

Dne 20. 11. 18 v 23:52 Kevin Fenzi napsal(a):
> On 11/20/18 2:39 PM, Frantisek Zatloukal wrote:
> > Hi,
> > I am currently trying to build new resultsdb[0] into epel7-candidate.
> > However, I am facing some (weird?) behavior around python-flask,
> >
> > I am able to do a successful build of resultsdb (depending on python-flask)
> > in mock epel-7 and koji scratch epel-7 [1]. However, it is failing on
> > not-available python-flask in koji to epel-7-candidate [2].
> >
> > Am I overlooking something?
> >
> > Thanks very much!
> >
> > [0] https://src.fedoraproject.org/rpms/resultsdb/blob/epel7/f/resultsdb.spec
> > [1] https://koji.fedoraproject.org/koji/taskinfo?taskID=31025152
> > [2] https://koji.fedoraproject.org/koji/taskinfo?taskID=31025187
>
> flask was pulled into rhel-extras (for docker distribution I think) and
> when it was, it was only published for x86_64.
>
> Your package is noarch, but flask is not.
>
> So, you happened to get a x86_64 builder with your scratch build, but
> your real build got a ppc builder where flask is not available. ;(
>
> I think you should be able to pass "Exclusivearch: noarch x86_64"

Was this resolved?

https://pagure.io/releng/issue/7671

V.
> and
> have it work. If not, you can just resubmit your build until you get a
> x86_64 builder.
>
> kevin
>
> infrastructure mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
> Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives: https://lists.fedoraproject.org/archives/list/[e-mail]...

[Kevin Fenzi]

On 11/21/18 3:11 AM, Vít Ondruch wrote:
> Dne 20. 11. 18 v 23:52 Kevin Fenzi napsal(a):
> > On 11/20/18 2:39 PM, Frantisek Zatloukal wrote:
> > > Hi,
> > > I am currently trying to build new resultsdb[0] into epel7-candidate.
> > > However, I am facing some (weird?) behavior around python-flask,
> > >
> > > I am able to do a successful build of resultsdb (depending on python-flask)
> > > in mock epel-7 and koji scratch epel-7 [1]. However, it is failing on
> > > not-available python-flask in koji to epel-7-candidate [2].
> > >
> > > Am I overlooking something?
> > >
> > > Thanks very much!
> > >
> > > [0] https://src.fedoraproject.org/rpms/resultsdb/blob/epel7/f/resultsdb.spec
> > > [1] https://koji.fedoraproject.org/koji/taskinfo?taskID=31025152
> > > [2] https://koji.fedoraproject.org/koji/taskinfo?taskID=31025187
> >
> > flask was pulled into rhel-extras (for docker distribution I think) and
> > when it was, it was only published for x86_64.
> >
> > Your package is noarch, but flask is not.
> >
> > So, you happened to get a x86_64 builder with your scratch build, but
> > your real build got a ppc builder where flask is not available. ;(
> >
> > I think you should be able to pass "Exclusivearch: noarch x86_64"
>
> Was this resolved?
>
> https://pagure.io/releng/issue/7671

I thought so, but the builds there have their logs cleaned up and the
devel spec no longer has this in it.
I tried a test package that I scratch built and it seemed to work fine.
(ie, when I had ExclusiveArch: noarch aarc64 it always went to a aarch64
builder).

Can you try again and let us know if you still hit something?

kevin

[Mikolaj Izdebski]

On Sat, Nov 24, 2018 at 8:08 PM Kevin Fenzi  wrote:
> On 11/21/18 3:11 AM, Vít Ondruch wrote:
> > Dne 20. 11. 18 v 23:52 Kevin Fenzi napsal(a):
> > Was this resolved?
> >
> > https://pagure.io/releng/issue/7671
>
> I thought so, but the builds there have their logs cleaned up and the
> devel spec no longer has this in it.
> I tried a test package that I scratch built and it seemed to work fine.
> (ie, when I had ExclusiveArch: noarch aarc64 it always went to a aarch64
> builder).

"ExclusiveArch: aarch64" works while "ExclusiveArch: x86_64" does not.
The problem is that Koji may try to build packages with
"ExclusiveArch: x86_64" in i386 chroot. This is not fixed AFAIK. A fix
would be splitting x86_64/i386 hosts into two distinct groups and
assigning exactly one arch to each builder.

--
Mikolaj

[Kevin Fenzi]

On 11/26/18 1:59 AM, Mikolaj Izdebski wrote:
> On Sat, Nov 24, 2018 at 8:08 PM Kevin Fenzi  wrote:
> > On 11/21/18 3:11 AM, Vít Ondruch wrote:
> > > Dne 20. 11. 18 v 23:52 Kevin Fenzi napsal(a):
> > > Was this resolved?
> > >
> > > https://pagure.io/releng/issue/7671
> >
> > I thought so, but the builds there have their logs cleaned up and the
> > devel spec no longer has this in it.
> > I tried a test package that I scratch built and it seemed to work fine.
> > (ie, when I had ExclusiveArch: noarch aarc64 it always went to a aarch64
> > builder).
>
> "ExclusiveArch: aarch64" works while "ExclusiveArch: x86_64" does not.
> The problem is that Koji may try to build packages with
> "ExclusiveArch: x86_64" in i386 chroot. This is not fixed AFAIK. A fix
> would be splitting x86_64/i386 hosts into two distinct groups and
> assigning exactly one arch to each builder.

That seems very odd to me. Do you have an example build of this?

Shouldn't mock pass the right thing here?

kevin

[Mikolaj Izdebski]

On Mon, Nov 26, 2018 at 8:17 PM Kevin Fenzi  wrote:
> On 11/26/18 1:59 AM, Mikolaj Izdebski wrote:
> > "ExclusiveArch: aarch64" works while "ExclusiveArch: x86_64" does not.
> > The problem is that Koji may try to build packages with
> > "ExclusiveArch: x86_64" in i386 chroot. This is not fixed AFAIK. A fix
> > would be splitting x86_64/i386 hosts into two distinct groups and
> > assigning exactly one arch to each builder.
>
> That seems very odd to me. Do you have an example build of this?

Example task: https://koji.stg.fedoraproject.org/koji/taskinfo?taskID=90003488
Package has "ExclusiveArch: x86_64", buildArch task has correct "Arch
x86_64", yet the build is ran in i386 chroot (see root.log) and fails
with "error: Architecture is not included: i386".
> Shouldn't mock pass the right thing here?

This is a bug in Koji, not mock.

--
Mikolaj

[Mikolaj Izdebski]

On Mon, Nov 26, 2018 at 11:37 PM Mikolaj Izdebski  wrote:
> On Mon, Nov 26, 2018 at 8:17 PM Kevin Fenzi  wrote:
> > On 11/26/18 1:59 AM, Mikolaj Izdebski wrote:
> > > "ExclusiveArch: aarch64" works while "ExclusiveArch: x86_64" does not.
> > > The problem is that Koji may try to build packages with
> > > "ExclusiveArch: x86_64" in i386 chroot. This is not fixed AFAIK. A fix
> > > would be splitting x86_64/i386 hosts into two distinct groups and
> > > assigning exactly one arch to each builder.
> >
> > That seems very odd to me. Do you have an example build of this?
>
> Example task: https://koji.stg.fedoraproject.org/koji/taskinfo?taskID=90003488
> Package has "ExclusiveArch: x86_64", buildArch task has correct "Arch
> x86_64", yet the build is ran in i386 chroot (see root.log) and fails
> with "error: Architecture is not included: i386".
> > Shouldn't mock pass the right thing here?
>
> This is a bug in Koji, not mock.

Reference to Koji issue: https://pagure.io/koji/issue/789

--
Mikołaj

[Adam Williamson]

On Tue, 2018-11-20 at 14:52 -0800, Kevin Fenzi wrote:
> On 11/20/18 2:39 PM, Frantisek Zatloukal wrote:
> > Hi,
> > I am currently trying to build new resultsdb[0] into epel7-candidate.
> > However, I am facing some (weird?) behavior around python-flask,
> >
> > I am able to do a successful build of resultsdb (depending on python-flask)
> > in mock epel-7 and koji scratch epel-7 [1]. However, it is failing on
> > not-available python-flask in koji to epel-7-candidate [2].
> >
> > Am I overlooking something?
> >
> > Thanks very much!
> >
> > [0] https://src.fedoraproject.org/rpms/resultsdb/blob/epel7/f/resultsdb.spec
> > [1] https://koji.fedoraproject.org/koji/taskinfo?taskID=31025152
> > [2] https://koji.fedoraproject.org/koji/taskinfo?taskID=31025187
>
> flask was pulled into rhel-extras (for docker distribution I think) and
> when it was, it was only published for x86_64.
>
> Your package is noarch, but flask is not.
>
> So, you happened to get a x86_64 builder with your scratch build, but
> your real build got a ppc builder where flask is not available. ;(
>
> I think you should be able to pass "Exclusivearch: noarch x86_64" and
> have it work. If not, you can just resubmit your build until you get a
> x86_64 builder.

There are several packages like this, I think. It's a real pain. Is
there no better solution we can come up with?