Hi Lars,<br><br>We think the model you are describing will be easily achieved in the upcoming 3.0 [1] release.<br>The approach we are taking to implement persistent VM is making the VM Template the main VM entity, as you pointed out.<br>
<br><br>With the new VM Template pool, you will be able to:<br><br><blockquote style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;" class="gmail_quote">(a) A vm instance would start with its own copy of a disk image.<br>
</blockquote><div><br>A new 'oneimage clone' command will provide a personal private copy of another Image. Then, a VM Template using that Image can be registered using 'onetemplate create my_vm.one'<br> <br>
</div><blockquote style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;" class="gmail_quote">
(b) Shutdown would not delete the image.<br></blockquote><div><br>The data won't be lost if the Image is persistent. Which I think is a reasonable default for an Image created from 'oneimage clone'.<br> </div>
<blockquote style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;" class="gmail_quote">
(c) Some command (restart, or resubmit, or something) would allow one<br>
to restore a vm instance that had been shut down.<br></blockquote><div><br>A VM instance will finish it's life-cycle after shutdown, this won't change. But the VM Template will be still available in the pool, and can be instantiated many more times.<br>
<br></div><blockquote style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;" class="gmail_quote">
(d) Disk data would only be discarded with an explicit "delete".<br></blockquote>
<br>There will a explicit delete for the disks (oneimage delete), and for the VM Template (onetemplate delete). <br><br><br>Best regards,<br>Carlos.<br><br>[1] <a href="http://blog.opennebula.org/?p=1626">http://blog.opennebula.org/?p=1626</a><br>
<br clear="all"><span style="border-collapse:collapse;color:rgb(136, 136, 136);font-family:arial, sans-serif;font-size:13px">--<br>Carlos Martín, MSc<br>Project Major Contributor<br><span style="background-color:rgb(255, 255, 204);color:rgb(34, 34, 34);background-repeat:initial initial">OpenNebula</span> - The Open Source Toolkit for Cloud Computing<br>
<a href="http://www.opennebula.org/" style="color:rgb(42, 93, 176)" target="_blank">www.<span style="background-color:rgb(255, 255, 204);color:rgb(34, 34, 34);background-repeat:initial initial">OpenNebula</span>.org</a> | <a href="mailto:cmartin@opennebula.org" style="color:rgb(42, 93, 176)" target="_blank">cmartin@<span style="background-color:rgb(255, 255, 204);color:rgb(34, 34, 34);background-repeat:initial initial">opennebula</span>.org</a></span><br>
<br><br><div class="gmail_quote">On Tue, May 24, 2011 at 10:08 PM, Lars Kellogg-Stedman <span dir="ltr"><<a href="mailto:lars@seas.harvard.edu" target="_blank">lars@seas.harvard.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Steffen, Tino,<br>
<br>
Thank you both for your answers. Using persistent disk images (or the<br>
"save as" feature) helps a little bit, but what I was really hoping<br>
for was a model in which:<br>
<br>
(a) A vm instance would start with its own copy of a disk image.<br>
(b) Shutdown would not delete the image.<br>
(c) Some command (restart, or resubmit, or something) would allow one<br>
to restore a vm instance that had been shut down.<br>
(d) Disk data would only be discarded with an explicit "delete".<br>
<br>
Ideally, this would be controlled at the virtual machine level, not<br>
the disk image level. That is, one would set the "don't delete on<br>
shutdown" option on the vm instance.<br>
<br>
The problem with "persistent" images is that -- as far as I can tell<br>
-- the vm instance references the image in the repository directly,<br>
rather than getting "private" persistent *copy* of the image. The<br>
problem with "saving" a vm instance is that it doesn't really address<br>
the problem of, "whoops, I typed the wrong vm id and shutdown the<br>
wrong system". It also forces the template description file to become<br>
more important, since the "recovery" process is to create a new<br>
virtual machine instance, rather than rebooting the old one.<br>
<br>
I have seen several similar questions on the list from folks looking<br>
for a way to support persistent virtual machines, so it looks as if<br>
this is a model for which there is some demand. For those of you<br>
familiar with the codebase, does this look as if it would be hard to<br>
implement? If I wanted to implement it myself, should I be looked at<br>
the 2.4 code? We're running 2.2 locally, but I understand that 2.4 is<br>
on the horizon.<br>
<font color="#888888"><br>
--<br>
</font><div>Lars Kellogg-Stedman <<a href="mailto:lars@seas.harvard.edu" target="_blank">lars@seas.harvard.edu</a>><br>
Senior Technologist<br>
Harvard University SEAS<br>
Academic and Research Computing (ARC)<br>
_______________________________________________<br>
</div><div><div></div><div>Users mailing list<br>
<a href="mailto:Users@lists.opennebula.org" target="_blank">Users@lists.opennebula.org</a><br>
<a href="http://lists.opennebula.org/listinfo.cgi/users-opennebula.org" target="_blank">http://lists.opennebula.org/listinfo.cgi/users-opennebula.org</a><br>
</div></div></blockquote></div><br>