Re: [openstack-dev] [ironic] [qa] Testing config drive creation in our CI

2016-09-26 Thread Jim Rollenhagen
On Mon, Sep 26, 2016 at 7:29 AM, Sean Dague  wrote:
> On 09/26/2016 07:15 AM, Dmitry Tantsur wrote:
>> On 09/26/2016 01:09 PM, Sean Dague wrote:
>>> This should probably be set at the job level, and not buried inside
>>> devstack to be different based on hypervisor. That's going to be a lot
>>> more confusing to unwind later.
>>
>> Fair. So should we just set DEVSTACK_GATE_CONFIGDRIVE for all our jobs?
>> Do you think it should go somewhere here:
>> https://github.com/openstack-infra/devstack-gate/blob/7ecc7dd4067d99e0fa7525a9fffc8b05e1a7b58f/devstack-vm-gate.sh#L343-L370
>> ?
>
> The devstack-gate change is probably sufficient.
>
> That being said, you can also add "config_drive": True to the server
> create json per request, and it will use a config drive. That may be a
> better option for testing, as it will let you specify at the test level
> what needs config drive testing.
>
> http://developer.openstack.org/api-ref/compute/?expanded=create-server-detail#id7

Well, we use some of the nova tests from tempest's trees, so that
probably isn't an option here. :)

We've been trying a bit to move things out of d-s-g and into project-config,
so I'd rather put it into project-config unless we have a good reason not to
do so.

// jim

__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


Re: [openstack-dev] [ironic] [qa] Testing config drive creation in our CI

2016-09-26 Thread Sean Dague
On 09/26/2016 07:15 AM, Dmitry Tantsur wrote:
> On 09/26/2016 01:09 PM, Sean Dague wrote:
>> This should probably be set at the job level, and not buried inside
>> devstack to be different based on hypervisor. That's going to be a lot
>> more confusing to unwind later.
> 
> Fair. So should we just set DEVSTACK_GATE_CONFIGDRIVE for all our jobs?
> Do you think it should go somewhere here:
> https://github.com/openstack-infra/devstack-gate/blob/7ecc7dd4067d99e0fa7525a9fffc8b05e1a7b58f/devstack-vm-gate.sh#L343-L370
> ?

The devstack-gate change is probably sufficient.

That being said, you can also add "config_drive": True to the server
create json per request, and it will use a config drive. That may be a
better option for testing, as it will let you specify at the test level
what needs config drive testing.

http://developer.openstack.org/api-ref/compute/?expanded=create-server-detail#id7

-Sean

-- 
Sean Dague
http://dague.net

__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


Re: [openstack-dev] [ironic] [qa] Testing config drive creation in our CI

2016-09-26 Thread Dmitry Tantsur

On 09/26/2016 01:09 PM, Sean Dague wrote:

This should probably be set at the job level, and not buried inside
devstack to be different based on hypervisor. That's going to be a lot
more confusing to unwind later.


Fair. So should we just set DEVSTACK_GATE_CONFIGDRIVE for all our jobs? Do you 
think it should go somewhere here: 
https://github.com/openstack-infra/devstack-gate/blob/7ecc7dd4067d99e0fa7525a9fffc8b05e1a7b58f/devstack-vm-gate.sh#L343-L370 
?




-Sean

On 09/26/2016 04:55 AM, Dmitry Tantsur wrote:

Just bringing QA folks attention: please merge
https://review.openstack.org/#/c/375467/ as we've regressed in our
testing coverage (see below for details).

On 09/23/2016 08:21 PM, Jim Rollenhagen wrote:

On Fri, Sep 23, 2016 at 7:37 AM, Dmitry Tantsur 
wrote:

Hi folks!

We've found out that we're not testing creating of config drives in
our CI.
It ended up in one combination being actually broken (pxe_* +
wholedisk +
configdrive). I would like to cover this testing gap. Is there any
benefit
in NOT using config drives in all jobs? I assume we should not bother
too
much testing the metadata service, as it's not within our code base
(unlike
config drive).

I've proposed https://review.openstack.org/375362 to switch our tempest
plugin to testing config drives, please vote. As you see one job
fails on it
- this is the breakage I was talking about. It will (hopefully) get
fixed
with the next release of ironic-lib.


Right, so as Pavlo mentioned in the patch, configdrive used to be the
default
for devstack, and as such we forced configdrive for all tests. When
that was
changed, we didn't notice because somehow metadata service worked.
https://github.com/openstack-dev/devstack/commit/7682ea88a6ab8693b215646f16748dbbc2476cc4


I agree, we should go back to using configdrive for all tests.

// jim



Finally, we need to run all jobs on ironic-lib, not only one, as
ironic-lib
is not the basis for all deployment variants. This will probably happen
after we switch our DSVM jobs to Xenial though.

-- Dmitry

__

OpenStack Development Mailing List (not for usage questions)
Unsubscribe:
openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


__

OpenStack Development Mailing List (not for usage questions)
Unsubscribe:
openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev




__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev






__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


Re: [openstack-dev] [ironic] [qa] Testing config drive creation in our CI

2016-09-26 Thread Sean Dague
This should probably be set at the job level, and not buried inside
devstack to be different based on hypervisor. That's going to be a lot
more confusing to unwind later.

-Sean

On 09/26/2016 04:55 AM, Dmitry Tantsur wrote:
> Just bringing QA folks attention: please merge
> https://review.openstack.org/#/c/375467/ as we've regressed in our
> testing coverage (see below for details).
> 
> On 09/23/2016 08:21 PM, Jim Rollenhagen wrote:
>> On Fri, Sep 23, 2016 at 7:37 AM, Dmitry Tantsur 
>> wrote:
>>> Hi folks!
>>>
>>> We've found out that we're not testing creating of config drives in
>>> our CI.
>>> It ended up in one combination being actually broken (pxe_* +
>>> wholedisk +
>>> configdrive). I would like to cover this testing gap. Is there any
>>> benefit
>>> in NOT using config drives in all jobs? I assume we should not bother
>>> too
>>> much testing the metadata service, as it's not within our code base
>>> (unlike
>>> config drive).
>>>
>>> I've proposed https://review.openstack.org/375362 to switch our tempest
>>> plugin to testing config drives, please vote. As you see one job
>>> fails on it
>>> - this is the breakage I was talking about. It will (hopefully) get
>>> fixed
>>> with the next release of ironic-lib.
>>
>> Right, so as Pavlo mentioned in the patch, configdrive used to be the
>> default
>> for devstack, and as such we forced configdrive for all tests. When
>> that was
>> changed, we didn't notice because somehow metadata service worked.
>> https://github.com/openstack-dev/devstack/commit/7682ea88a6ab8693b215646f16748dbbc2476cc4
>>
>>
>> I agree, we should go back to using configdrive for all tests.
>>
>> // jim
>>
>>>
>>> Finally, we need to run all jobs on ironic-lib, not only one, as
>>> ironic-lib
>>> is not the basis for all deployment variants. This will probably happen
>>> after we switch our DSVM jobs to Xenial though.
>>>
>>> -- Dmitry
>>>
>>> __
>>>
>>> OpenStack Development Mailing List (not for usage questions)
>>> Unsubscribe:
>>> openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>
>> __
>>
>> OpenStack Development Mailing List (not for usage questions)
>> Unsubscribe:
>> openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>
> 
> 
> __
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


-- 
Sean Dague
http://dague.net

__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


Re: [openstack-dev] [ironic] [qa] Testing config drive creation in our CI

2016-09-26 Thread Dmitry Tantsur
Just bringing QA folks attention: please merge 
https://review.openstack.org/#/c/375467/ as we've regressed in our testing 
coverage (see below for details).


On 09/23/2016 08:21 PM, Jim Rollenhagen wrote:

On Fri, Sep 23, 2016 at 7:37 AM, Dmitry Tantsur  wrote:

Hi folks!

We've found out that we're not testing creating of config drives in our CI.
It ended up in one combination being actually broken (pxe_* + wholedisk +
configdrive). I would like to cover this testing gap. Is there any benefit
in NOT using config drives in all jobs? I assume we should not bother too
much testing the metadata service, as it's not within our code base (unlike
config drive).

I've proposed https://review.openstack.org/375362 to switch our tempest
plugin to testing config drives, please vote. As you see one job fails on it
- this is the breakage I was talking about. It will (hopefully) get fixed
with the next release of ironic-lib.


Right, so as Pavlo mentioned in the patch, configdrive used to be the default
for devstack, and as such we forced configdrive for all tests. When that was
changed, we didn't notice because somehow metadata service worked.
https://github.com/openstack-dev/devstack/commit/7682ea88a6ab8693b215646f16748dbbc2476cc4

I agree, we should go back to using configdrive for all tests.

// jim



Finally, we need to run all jobs on ironic-lib, not only one, as ironic-lib
is not the basis for all deployment variants. This will probably happen
after we switch our DSVM jobs to Xenial though.

-- Dmitry

__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev


__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev




__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev