fuel-dev team mailing list archive

[Dmitriy Shulyak <dshulyak@xxxxxxxxxxxx>]

+1 to the idea of separate handler for astute and "prepare" task in nailgun
if we want this in 6.0.

In future, i think, nailgun will simply skip generation of any deployment
tasks, if "prepare" api call requested.
If deployment will be described as set of tasks with pre/post-deployment
actions and will look like:

- upload_facts
- sync_puppet
- deploy_nova
- deploy_ceph

nailgun will omit deploy_nova, deploy_ceph tasks and execute only facts and
sync



On Thu, Oct 2, 2014 at 1:07 PM, Evgeniy L <eli@xxxxxxxxxxxx> wrote:

> Hi Vladimir,
>
> I think the way how it should be implemented needs to be discussed with
> Dmitry Shulyak
> who is working on Granular deployment.
> If we will chose to hardcode the logic on orchestrator side, we can have
> the problems
> with tasks reimplementation of this feature.
> From my point of view it should be separate task (or handler in terms of
> orchestrator), it
> should not be deployment task, which behaves differently if some flag in
> data is set as True.
>
> Thanks,
>
> On Wed, Oct 1, 2014 at 3:44 PM, Vladimir Sharshov <vsharshov@xxxxxxxxxxxx>
> wrote:
>
>> Hello.
>>
>> For development and debugging propose it would be nice if Fuel can
>> perform all deployment steps without actually deploy it:
>>
>>    1. provisioning;
>>    2. pre deployment steps (sync repositories and puppet modules,
>>    generate and upload ssh keys, sync time);
>>    3. upload Astute.yaml for each role (and support multi roles case).
>>
>> As result we will get cluster which ready for debugging and development
>> proposes.
>>
>> Possible solution:
>>
>>    - add new cluster parameter "dry_run" like "debug";
>>    - run all pre deploy and deployment steps;
>>    - upload astute.yaml for each role separately (/etc/<role>.yaml) and
>>    do symbolic to /etc/astute.yaml (
>>    https://review.openstack.org/#/c/124728/)
>>
>> Current workflow:
>>
>> We can do it by hands using fuel-cli or UI, but this step require to
>> spoil site.pp. Also without https://review.openstack.org/#/c/124728/ we
>> loosing multi roles support and this workaround will not work with granular
>> deployment because no single site.pp exists.
>>
>> Unresolved issues:
>>
>>    - 'dry run' in web UI;
>>    - final status for node and cluster in case of dry run deployment;
>>    - small separate action in fuel-cli available for provisioned nodes
>>    only or as special parameter for deploy action like debug.
>>
>>
>> --
>> Mailing list: https://launchpad.net/~fuel-dev
>> Post to     : fuel-dev@xxxxxxxxxxxxxxxxxxx
>> Unsubscribe : https://launchpad.net/~fuel-dev
>> More help   : https://help.launchpad.net/ListHelp
>>
>>
>
> --
> Mailing list: https://launchpad.net/~fuel-dev
> Post to     : fuel-dev@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~fuel-dev
> More help   : https://help.launchpad.net/ListHelp
>
>