[Openstack-operators] flavor management

Joe Topjian joe.topjian at cybera.ca
Tue Apr 2 17:15:21 UTC 2013


On Tue, Apr 2, 2013 at 10:48 AM, Joe Topjian <joe.topjian at cybera.ca> wrote:

>
>
>
> On Tue, Apr 2, 2013 at 9:15 AM, Joe Topjian <joe.topjian at cybera.ca> wrote:
>
>> Hi Jon,
>>
>> On Tue, Apr 2, 2013 at 8:47 AM, Jonathan Proulx <jon at jonproulx.com>wrote:
>>
>>> It would be nice to do per project flavors, wouldn't it...
>>>
>>> Currently all our cloud hardware is homogeneous so we divide available
>>> resources into proportional chunks, so mcuh memory and ephemeral disk
>>> (provided from node local storage) per vCPU.  It's rather simplistic and
>>> still heavily influenced by the default flavors (all have a 10G root with
>>> the rest of the block storage  as an ephemeral vdb)
>>>
>>> We do have one flavor with a 40G root named "m1.windows" but don't
>>> provide a public windows image and as far as I know noone but
>>> administrators in my group have used it for testing (not a big windows
>>> demand here).
>>>
>>
>> Interesting. We had a "w1.large" flavor for a short period but ended up
>> removing it because we wanted all flavors to be able to work with all
>> images.
>>
>>
>>>
>>> On Mon, Apr 1, 2013 at 6:22 PM, Joe Topjian <joe.topjian at cybera.ca>wrote:
>>>
>>>>
>>>>
>>>> Boot from Volume might be a possibility, but does anyone have a
>>>> procedure to create a Windows bootable volume?
>>>>
>>>>
>>> With the large caveat that I don't have any Windows instances I do have
>>> a thought on where I'd start, given there's been no other response (on list
>>> anyway)
>>>
>>> 1st cinder is meant to be able to create volumes based on glance images
>>> (--image-id <image-id>), which would be delightful, but hasn't worked for
>>> me and I haven't had a chance to look at why, so unsure if this is a local
>>> issue for me or a bug.
>>>
>>
>> Thanks for the tip. I'm going to give this a shot.
>>
>
> This worked for me.
>

Spoke too soon. This does indeed work for standard LVM/iSCSI
configurations, like how my test environment is set up, but it is failing
immediately when using the NetApp driver. The error I am getting is:

OSError: [Errno 2] No such file or directory:
'/dev/mapper/cinder--volumes-volume--2c8e5316--5a94--4c7a--9ef7--0663c49feecf'

>From what I can tell, there isn't a proper local_path() function for iSCSI
and/or NetApp and so Cinder is falling back to the default driver's
local_path, which is LVM. Opening a bug now.


> The only change I had to make was adding "glance_host=" to my cinder.conf
> file. There are some other glance options (
> https://github.com/openstack/cinder/blob/master/etc/cinder/cinder.conf.sample#L92-L111)
> but once glance_host was set, the defaults were applicable to my
> environment.
>
> Next, I created a volume using the --image-id flag:
>
> cinder create --image-id=a9b4535f-c56e-4816-b6aa-a3caef7dbba4
> --display-name=win2k8r2vol 30
>
> Does anyone know off of the top of their head if this is possible to do
> through Horizon in Grizzly?
>
> Finally, I launched an instance via Horizon. The only clunky thing was
> having to specify an image even though it's ignored. The docs correctly
> mention that this is a bug (
> http://docs.openstack.org/trunk/openstack-compute/admin/content/boot-from-volume.html),
> but it's supposed to have been fixed in Grizzly. Does anyone have a Grizzly
> environment to test this out and confirm it's fixed (especially in
> Horizon)? If not, I'll try to do this relatively soon.
>
>
>
>>
>>
>>>
>>> Failing that I'd create the volume in cinder then attach it directly to
>>> a linux host somewhere and either run the install directly using
>>> appropriate virtualization tools or grab an existing windows image from
>>> glance (either just by grabbing the file or using glance image-download
>>> depending on your setup)
>>>
>>> -Jon
>>>
>>
>>
>>
>> --
>> Joe Topjian
>> Systems Administrator
>> Cybera Inc.
>>
>> www.cybera.ca
>>
>> Cybera is a not-for-profit organization that works to spur and support
>> innovation, for the economic benefit of Alberta, through the use
>> of cyberinfrastructure.
>>
>
>
>
> --
> Joe Topjian
> Systems Administrator
> Cybera Inc.
>
> www.cybera.ca
>
> Cybera is a not-for-profit organization that works to spur and support
> innovation, for the economic benefit of Alberta, through the use
> of cyberinfrastructure.
>



-- 
Joe Topjian
Systems Administrator
Cybera Inc.

www.cybera.ca

Cybera is a not-for-profit organization that works to spur and support
innovation, for the economic benefit of Alberta, through the use
of cyberinfrastructure.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-operators/attachments/20130402/070dfa4c/attachment.html>


More information about the OpenStack-operators mailing list