fuel-dev team mailing list archive

[Vladimir Kuklin <vkuklin@xxxxxxxxxxxx>]

Ryan, we are running the same classes and configuring several parameters
that can intersect with controller node, e.g. "enabled_apis" nova.conf
parameter is different for compute and controller node. In case you run
compute installation after the controller is installed, you get into state
when "enabled_apis" equals to "metadata" and nova-api restart stops all
other services(nova-api for example) leaving only metadata service running.


On Tue, Mar 11, 2014 at 11:30 PM, David Easter <deaster@xxxxxxxxxxxx> wrote:

> I will say that if we can support an "all in one" installation, that would
> be a definite plus.
>
> So as long as there aren't any negative consequences, I'd give this a +1.
>
> Thanks,
>
> - David J. Easter
>   Product Line Manager, Mirantis
>
> From: Ryan Moe <rmoe@xxxxxxxxxxxx>
> Date: Tuesday, March 11, 2014 at 12:28 PM
> To: Vladimir Kuklin <vkuklin@xxxxxxxxxxxx>
> Cc: fuel-dev <fuel-dev@xxxxxxxxxxxxxxxxxxx>
> Subject: Re: [Fuel-dev] Combined controller and compute node
>
> What kind of side stuff are we doing and what kind of problems will it
> lead to?
>
> On Tue, Mar 11, 2014 at 12:20 PM, Vladimir Kuklin <vkuklin@xxxxxxxxxxxx>wrote:
>
>> Ryan
>>
>> There could be some intersections between controller and compute node
>> configuration. I am still not quite sure that installing controller and
>> compute configs will not break anything: you will need to run tempest at
>> least. We are running 2 contradicting classes, which we should not.
>> Instead, we should split deployment into small pieces and install only
>> nova-compute and dependencies on the controller node. Currently, we are
>> doing a lot if side stuff, that can lead us to a lot of problems.
>>
>>
>> On Tue, Mar 11, 2014 at 11:05 PM, Mike Scherbakov <
>> mscherbakov@xxxxxxxxxxxx> wrote:
>>
>>> Great finding!
>>> I believe there were some minor issues, and there were no use cases for
>>> using all-in-one deployment. That's why we kept it as is, and didn't even
>>> bother fixing the issues. If we remove restriction, then it means we have
>>> to test it thoroughly. For this particular case, there is 95% guarantee
>>> that it will work just fine on one node if it works on two.
>>>
>>> So I'm +1 for removing the restriction. It will also allow us to rewrite
>>> some smoke tests to use single-node env to speed up Fuel CI checks.
>>>
>>> However we must keep at least system tests for distributed envs, as it
>>> is easy to hardcode "localhost" somewhere and we must always ensure that
>>> Fuel works in distributed mode.
>>>
>>>
>>> On Tue, Mar 11, 2014 at 10:57 PM, Ryan Moe <rmoe@xxxxxxxxxxxx> wrote:
>>>
>>>> What were the reasons for not allowing a node to have both the compute
>>>> and controller roles? In my initial testing it appears that this works
>>>> fine. OSTF passes, all controller functions continue working, VMs boot,
>>>> etc. If those original concerns are no longer valid or if they can be
>>>> addressed then we can remove another arbitrary restriction from Fuel.
>>>>
>>>> Thoughts?
>>>>
>>>> -Ryan
>>>>
>>>> --
>>>> 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
>>>>
>>>>
>>>
>>>
>>> --
>>> Mike Scherbakov
>>> #mihgen
>>>
>>> --
>>> 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
>>>
>>>
>>
>>
>> --
>> Yours Faithfully,
>> Vladimir Kuklin,
>> Senior Deployment Engineer,
>> Mirantis, Inc.
>> +7 (495) 640-49-04
>> +7 (926) 702-39-68
>> Skype kuklinvv
>> 45bk3, Vorontsovskaya Str.
>> Moscow, Russia,
>> www.mirantis.com <http://www.mirantis.ru/>
>> www.mirantis.ru
>> vkuklin@xxxxxxxxxxxx
>>
>
> -- Mailing list: https://launchpad.net/~fuel-dev Post to :
> fuel-dev@xxxxxxxxxxxxxxxxxxx Unsubscribe : https://launchpad.net/~fuel-devMore help :
> https://help.launchpad.net/ListHelp
>



-- 
Yours Faithfully,
Vladimir Kuklin,
Senior Deployment Engineer,
Mirantis, Inc.
+7 (495) 640-49-04
+7 (926) 702-39-68
Skype kuklinvv
45bk3, Vorontsovskaya Str.
Moscow, Russia,
www.mirantis.com <http://www.mirantis.ru/>
www.mirantis.ru
vkuklin@xxxxxxxxxxxx