[one-users] Writing one template that allows for persistent VM:

[Steven Timm]

hOn Thu, 3 Feb 2011, Steven Timm wrote:

>
> I am trying to come up with a prescription for users that allows the
> following sequence:
>
> 1) Launch a VM based on a standard VM repository template
>
> 2) Save the contents of the VM back to the VM repositor\
>
> 3) Relaunch the VM without modifying the template.
>
> I've tried the following thus far:
>
> 1) Launch the VM initially from a VM OS file outside the VM repository
> with save=yes.  this does save the VM OS but just to
> /var/lib/one/<vmid>/disk.0, not back to the original file.
> User must copy that file somewhere and/or change their template.
>
> 1a) Launch the VM and then execute a onevm saveas command.
>    this gets a copy of the image back to the image repository
>    but again user must modify the template to relaunch the VM.
>
> 2) Make a public VM image in the repository, launch from that.
> Problems--even if image is public, other users (even oneadmin) can't read it 
> and sometimes I myself can't read it, without going into
> the image repo and chmod'ing the file by hand.  And again
> you are forced to save back to something other than the original file.
>
> 3) Make a persistent non-public image in the repository.
> Again there are permission problems launching, and it can only
> be used once by one user.  Even so I can't get it to save back to the
> repository.
>
If the user makes and saves the persistent non-public image
in the repository, it is then necessary to go in and
change the permissions on the file in the repository, otherwise
the oneadmin user can't read it.  But once this is done,
it is possible to start the VM from the repository and save it
back to the same place in the repository when it is done.

Next step--to make a public non-persistent generic image
in the repo, and then have a "clone" function such that
a user can make a non-public persistent image from it and
fork off their VM and save it from there on.  Anyone else interested?

Steve Timm




> Everything that we need to do is doable, but it takes hacking around
> with the template every time to be able to launch where we
> left off.  We need a way to make things more seamless.
>
> Steve Timm
>
>
>
>

-- 
------------------------------------------------------------------
Steven C. Timm, Ph.D  (630) 840-8525
timm at fnal.gov  http://home.fnal.gov/~timm/
Fermilab Computing Division, Scientific Computing Facilities,
Grid Facilities Department, FermiGrid Services Group, Group Leader.
Lead of FermiCloud project.