openstack team mailing list archive
-
openstack team
-
Mailing list archive
-
Message #18636
Re: Controller node vs. compute node terminology
-
To:
Anne Gentle <anne@xxxxxxxxxxxxx>
-
From:
Ahmed Al-Mehdi <ahmed@xxxxxxxxxx>
-
Date:
Thu, 15 Nov 2012 14:18:36 -0600
-
Accept-language:
en-US
-
Acceptlanguage:
en-US
-
Cc:
"openstack@xxxxxxxxxxxxxxxxxxx" <openstack@xxxxxxxxxxxxxxxxxxx>
-
In-reply-to:
<CAD0KtVHHvYv3=HJpJLeWng9ybt1sEw8saaBT3gxyV--z3796bw@mail.gmail.com>
-
Thread-index:
Ac3DbmSEuGzpafPcRQK/SBH8JTsL/g==
-
Thread-topic:
[Openstack] Controller node vs. compute node terminology
-
User-agent:
Microsoft-MacOutlook/14.2.4.120824
Hi Anne,
Thank you very much for the clarification. I was thinking along the lines
of your explanation (since nova-compute is being installed on the
controller node per the doc), however your explanation solidified my
understanding.
As a suggestion, it would be good to incorporate this info into the
document (in some shape or form), so it is explicitly stated.
Just throwing out a suggestion regarding the doc, no disrespect to the
authors of the manual. My feeling is the all-sevices-on-one-node would
not be a typical deployment for "any" clouds (my opinion). A user who
wants to seriously evaluate/learn about OpenStack would start out with two
servers, one being a "pure" control node and other being a compute node.
This would be closer to a "typical" deployment and would increase
understanding of OpenStack (networking aspect, which is very critical).
Would it be more beneficial for the manual to document such a setup (as
opposed to an all-sevices-on-one-node server, plus additional compute
nodes).
Regards,
Ahmed.
On 11/15/12 11:49 AM, "Anne Gentle" <anne@xxxxxxxxxxxxx> wrote:
>On Thu, Nov 15, 2012 at 1:40 PM, Ahmed Al-Mehdi <ahmed@xxxxxxxxxx> wrote:
>> Is your statement based on the setup noted in the "OpenStack Install and
>> Deploy Manual"? I am question was more specific to the setup covered
>>in
>> the afore mentioned document.
>
>Specific to that guide, the controller node also contains
>nova-compute. Any node with that service (nova-compute) can act as a
>Compute node and launch VMs.
>
>This all-services-on-one-node setup on the controller node is not a
>typical deployment for large-scale clouds (public or private).
>
>> In general, does the answer not depend on what services are installed
>>on the
>> controller node, to give it "ability" to be a compute node also?
>
>The services installed do define its abilities, yes. Often a
>"Controller node" contains items like databases (such as MySQL),
>messaging queues (such as Rabbit-MQ), identity, images, all
>nova-services minus nova-compute, volumes, and the dashboard. Often a
>"Compute node" offers the hypervisor, nova-compute, the API metadata
>service (nova-api-metadata) and possibly the networking services,
>either nova-network or Quantum.
>
>You could also review the "Basic install guide" at
>https://review.openstack.org/#/c/16096/ which outlines it quite
>prescriptively.
>
>In your environment you may have reasons for splitting out services
>differently, how that division is done depends on your use case. The
>"Controller" and "Compute" nodes are an easy nomenclature and pattern
>for many cases.
>
>Hope this helps,
>Anne
>
>> Regards,
>> Ahmed.
>>
>>
>> From: Pengjun Pan <panpengjun@xxxxxxxxx>
>> Date: Thursday, November 15, 2012 11:58 AM
>> To: Ahmed Al-Mehdi <ahmed@xxxxxxxxxx>
>> Cc: "openstack@xxxxxxxxxxxxxxxxxxx" <openstack@xxxxxxxxxxxxxxxxxxx>
>> Subject: Re: [Openstack] Controller node vs. compute node terminology
>>
>> The controller node can not host vms. But an All-in-one node can,
>>because it
>> is a combination of controller and compute.
>>
>>
>> On Thu, Nov 15, 2012 at 12:53 PM, Ahmed Al-Mehdi <ahmed@xxxxxxxxxx>
>>wrote:
>>>
>>> Hello,
>>>
>>> I am following the document "OpenStack Deploy and Install Guide" to
>>>bring
>>> up a OpenStack setup. Per the instructions in the document, is the
>>> controller node also acting as the compute node. By that I mean when
>>>a VM
>>> is created, it created on the controller node. Is that right? And
>>>then
>>> there are steps to create additional (sole) compute nodes. Is that a
>>> correct view of looking at the setup described in the document.
>>>
>>> Server 1 : Controller node / Compute node
>>> Server 2: Compute node
>>>
>>> If so, is that the usual deployment scenario?
>>>
>>> Any thoughts would be highly appreciated.
>>>
>>> Regards,
>>> Ahmed.
>>>
>>>
>>> _______________________________________________
>>> Mailing list: https://launchpad.net/~openstack
>>> Post to : openstack@xxxxxxxxxxxxxxxxxxx
>>> Unsubscribe : https://launchpad.net/~openstack
>>> More help : https://help.launchpad.net/ListHelp
>>>
>>
>>
>> _______________________________________________
>> Mailing list: https://launchpad.net/~openstack
>> Post to : openstack@xxxxxxxxxxxxxxxxxxx
>> Unsubscribe : https://launchpad.net/~openstack
>> More help : https://help.launchpad.net/ListHelp
>>
Follow ups
References