GCE - cloud init, or google-daemon/google-startup-scripts

[Matthew Miller]

The documentation for building an image for Google Compute Engine (GCE):
 
  https://developers.google.com/compute/docs/building-image

Most of this is very similar to our existing cloud images. A few of the
changes we could integrate across the board. Some will need testing
(hostname setting?), but I don't see anything exotic.

Debian is carrying a patch for cloud-init provided by Google 
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=725384, and we could do the
same instead of integrating google-daemon and google-startup-scripts. I
assume this will get upstream at some point (there is some discussion of
that in the Debian bug).

I don't _absolutely_ love cloud-init, but it's well-understood and common,
and fits with our existing images, making it easier to document how to do
things with our image, and etc. 

What do we think?


-- 
Matthew Miller  ☁☁☁  Fedora Cloud Architect  ☁☁☁  <mattdm(a)fedoraproject.org&gt;

[Lars Kellogg-Stedman]

On Wed, Nov 13, 2013 at 12:55:39PM -0500, Matthew Miller wrote:
>  I don't _absolutely_ love cloud-init, but it's
> well-understood and common,
>  and fits with our existing images, making it easier to document how to do
>  things with our image, and etc. 
>  
>  What do we think? 
I think the more places on which we can use the same initial
configuration mechanism the better.  If cloud-init can abstract out
the difference between AWS and Google such that I can create a single
image that will work correctly in both places, that seems awesome.

-- 
Lars Kellogg-Stedman <lars(a)redhat.com&gt; | larsks @ irc
Cloud Engineering / OpenStack          | "   "  @ twitter

[Haïkel]

Hi,

the more predictible for the user, the better.
Cloud-init does a great job abstracting image initialization accross 
platforms, so i'd rather stick to one method in our default image.

Anyway, google-daemon/initscript should be packaged and made available 
in Fedora anyway.

best regards,
H.

[Sam Kottler]

On 11/13/2013 12:55 PM, Matthew Miller wrote:
>  The documentation for building an image for Google Compute Engine
> (GCE):
>   
>    https://developers.google.com/compute/docs/building-image
>  
>  Most of this is very similar to our existing cloud images. A few of the
>  changes we could integrate across the board. Some will need testing
>  (hostname setting?), but I don't see anything exotic.
>  
>  Debian is carrying a patch for cloud-init provided by Google 
>  http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=725384, and we could do the
>  same instead of integrating google-daemon and google-startup-scripts. I
>  assume this will get upstream at some point (there is some discussion of
>  that in the Debian bug).
>  
>  I don't _absolutely_ love cloud-init, but it's well-understood and common,
>  and fits with our existing images, making it easier to document how to do
>  things with our image, and etc.  
+1 to sticking with cloud-init going forward. As I said during the
meeting, I'm not a huge fan, but it's predictable and modular. Also, up
to this point we've been trying to keep things similar across different
platforms and using cloud-init on GCE would help continue that pattern.

>  
>  What do we think?
>  
>

[Sandro "red" Mathys]

On Wed, Nov 13, 2013 at 7:13 PM, Sam Kottler <skottler(a)redhat.com&gt; wrote:
>  On 11/13/2013 12:55 PM, Matthew Miller wrote:
> > The documentation for building an image for Google Compute Engine (GCE):
> >
> >   https://developers.google.com/compute/docs/building-image
> >
> > Most of this is very similar to our existing cloud images. A few of the
> > changes we could integrate across the board. Some will need testing
> > (hostname setting?), but I don't see anything exotic.
> >
> > Debian is carrying a patch for cloud-init provided by Google
> > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=725384, and we could do the
> > same instead of integrating google-daemon and google-startup-scripts. I
> > assume this will get upstream at some point (there is some discussion of
> > that in the Debian bug).
> >
> > I don't _absolutely_ love cloud-init, but it's well-understood and common,
> > and fits with our existing images, making it easier to document how to do
> > things with our image, and etc.
>
>  +1 to sticking with cloud-init going forward. As I said during the
>  meeting, I'm not a huge fan, but it's predictable and modular. Also, up
>  to this point we've been trying to keep things similar across different
>  platforms and using cloud-init on GCE would help continue that pattern. 
+1 on everything Sam said. cloud-init isn't great but it's probably
the best universal approach out there and should be universally used
as such. Also, the (non-Fedora-specific) cloud user community has
grown accustomed to it by now.

> >
> > What do we think?
> >
> >
>
>  _______________________________________________
>  cloud mailing list
>  cloud(a)lists.fedoraproject.org
>  https://admin.fedoraproject.org/mailman/listinfo/cloud
>  Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct

[Frankie]

On Thu, Nov 14, 2013 at 9:55 AM, Sandro "red" Mathys
<red(a)fedoraproject.org&gt;wrote:

>  On Wed, Nov 13, 2013 at 7:13 PM, Sam Kottler
> <skottler(a)redhat.com&gt; wrote:
>  > On 11/13/2013 12:55 PM, Matthew Miller wrote:
>  >> The documentation for building an image for Google Compute Engine (GCE):
>  >>
>  >>   https://developers.google.com/compute/docs/building-image
>  >>
>  >> Most of this is very similar to our existing cloud images. A few of the
>  >> changes we could integrate across the board. Some will need testing
>  >> (hostname setting?), but I don't see anything exotic.
>  >>
>  >> Debian is carrying a patch for cloud-init provided by Google
>  >> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=725384, and we could
>  do the
>  >> same instead of integrating google-daemon and google-startup-scripts. I
>  >> assume this will get upstream at some point (there is some discussion of
>  >> that in the Debian bug).
>  >>
>  >> I don't _absolutely_ love cloud-init, but it's well-understood and
>  common,
>  >> and fits with our existing images, making it easier to document how to
>  do
>  >> things with our image, and etc.
>  >
>  > +1 to sticking with cloud-init going forward. As I said during the
>  > meeting, I'm not a huge fan, but it's predictable and modular. Also, up
>  > to this point we've been trying to keep things similar across different
>  > platforms and using cloud-init on GCE would help continue that pattern.
>
>  +1 on everything Sam said. cloud-init isn't great but it's probably
>  the best universal approach out there and should be universally used
>  as such. Also, the (non-Fedora-specific) cloud user community has
>  grown accustomed to it by now.
>  
+1 is all I have to say here. This is exactly what I think.
I will however give the Google tools a shot and see what next.

>  >>
>  >> What do we think?
>  >>
>  >>
>  >
>  > _______________________________________________
>  > cloud mailing list
>  > cloud(a)lists.fedoraproject.org
>  > https://admin.fedoraproject.org/mailman/listinfo/cloud
>  > Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
>  _______________________________________________
>  cloud mailing list
>  cloud(a)lists.fedoraproject.org
>  https://admin.fedoraproject.org/mailman/listinfo/cloud
>  Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
>  


-- 
Skype: Frankie.Onuonga
twitter: Frankieonuonga
irc #freenode: Frankie.onuonga