<div dir="ltr">I agree with Michael. We need to use the OpenStack tooling.<div><br></div><div>Sahara is encountering some of the same issues we are as they are building up their hadoop VM/clusters.</div><div><br></div><div>
See</div><div><br></div><div><a href="http://docs.openstack.org/developer/sahara/userdoc/vanilla_plugin.html">http://docs.openstack.org/developer/sahara/userdoc/vanilla_plugin.html</a><br></div><div><a href="http://docs.openstack.org/developer/sahara/userdoc/diskimagebuilder.html">http://docs.openstack.org/developer/sahara/userdoc/diskimagebuilder.html</a><br>
</div><div><br></div><div>for inspiration,</div><div><br></div><div>Susanne</div><div><br></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Wed, Aug 27, 2014 at 6:21 PM, Michael Johnson <span dir="ltr"><<a href="mailto:johnsomor@gmail.com" target="_blank">johnsomor@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">I am investigating building scripts that use diskimage-builder<br>
(<a href="https://github.com/openstack/diskimage-builder" target="_blank">https://github.com/openstack/diskimage-builder</a>) to create a "purpose<br>
built" image. This should allow some flexibility in the base image<br>
and the output image format (including a path to docker).<br>
<br>
The definition of "purpose built" is open at this point. I will<br>
likely try to have a minimal Ubuntu based VM image as a starting<br>
point/test case and we can add/change as necessary.<br>
<br>
Michael<br>
<div><div class="h5"><br>
<br>
On Wed, Aug 27, 2014 at 2:12 PM, Dustin Lundquist <<a href="mailto:dustin@null-ptr.net">dustin@null-ptr.net</a>> wrote:<br>
> It seems to me there are two major approaches to the Octavia VM design:<br>
><br>
> Start with a standard Linux distribution (e.g. Ubuntu 14.04 LTS) and install<br>
> HAProxy 1.5 and Octavia control layer<br>
> Develop a minimal purpose driven distribution (similar to m0n0wall) with<br>
> just HAProxy, iproute2 and a Python runtime for the control layer.<br>
><br>
> The primary difference here is additional development effort for option 2,<br>
> verses the increased image size of option 1. Using Ubuntu and CirrOS images<br>
> a representative of the two options it looks like the image size difference<br>
> is on the about 20 times larger for a full featured distribution. If one of<br>
> the HA models is to spin up a replacement instance on failure the image size<br>
> could be significantly affect fail-over time.<br>
><br>
> For initial work I think starting with a standard distribution would be<br>
> sensible, but we should target systemd (Debian adopted systemd as new<br>
> default, and Ubuntu is following suit). I wanted to find out if there is<br>
> interest in a minimal Octavia image, and if so this may affect design<br>
> decisions on the instance control plane component.<br>
><br>
><br>
> -Dustin<br>
><br>
</div></div>> _______________________________________________<br>
> OpenStack-dev mailing list<br>
> <a href="mailto:OpenStack-dev@lists.openstack.org">OpenStack-dev@lists.openstack.org</a><br>
> <a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev" target="_blank">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a><br>
><br>
<br>
_______________________________________________<br>
OpenStack-dev mailing list<br>
<a href="mailto:OpenStack-dev@lists.openstack.org">OpenStack-dev@lists.openstack.org</a><br>
<a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev" target="_blank">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a><br>
</blockquote></div><br></div>