Feature request: copied from bug 1812901. virtualbox power type (or equiv hooks)


#1

This is a fairly absurd process. Just because the team has decided that features should no longer be submitted via the bug process doesn’t mean the user should have to redo the work (they could have been automatically transferred).

Alberto Donato 1812901@bugs.launchpad.net via canonical.com
4:31 AM (6 hours ago)
to me

Thanks for your feature request, please could you start a thread over on
our Discourse (https://discourse.maas.io/c/features) to discuss the
feature?

** Changed in: maas
Status: Opinion => Invalid

<“invalid:” is an absurd translation of RFE">


You received this bug notification because you are subscribed to the bug
report.
https://bugs.launchpad.net/bugs/1812901

Title:
power type virtualbox

Status in MAAS:
Invalid

Bug description:
Request for Enhancement

blog.ubuntu.com/2015/01/15/virtualbox-extensions-for-maas has a lovely
writeup explaining motivation and a Mac OS X hosted VirtualBox/MAAS
mashup. Unfortunately, the hooks it uses no longer seem to exist in
modern MAAS releases. As all of the bits are stale (Mac OS X,
VirtualBox, MAAS) I sought a modern replacement. In MAAS 2.4 and 2.5,
there is “virsh” power type support, which ought to suffice. However,
the current Homebrew formulae fail (obviously, NOT canonical’s problem
or fault). However, it would be nice if MAAS “stood on it’s own”, and
simply provided direct support for a VirtualBox powertype. If that
isn’t something the project is willing to provide (VboxManage calls
should provide sufficient functionality) then providing a generic set
of scripts which will be invoked by MAAS would enable a smooth path
for users.

Since there’s virsh, vmware and openstack powertypes, it seems like
this ought to be within reason.

There’s no obvious reason to not handle Parallels, but it’s been some
years since I’ve used it.

Adding a pointer to a newer writeup in the existing blog post would be
helpful.

In the alternative, leveraging the Apple xhyve hypervisor might be
useful … but I suspect that for fullest fidelity to a physical
deployment, the VirtualBox approach may be most helpful.

My plan was to develop a field training module … so technicians
could get some hands on before hacking on the actual production
hardware. VirtualBox would be ideal in that it would work on Mac and
Windows removing potential impediments to deployment.

My management has reorganized the team somewhat; the folks who currently are focused on the aspects of our Edge compute infrastructure may (or may not) have moved away from MAAS as their approach of choice. However, if I were still involved, a Lab/introduction course based on VirtualBox would have been a must have. The folks that would need to be trained seldom have ample physical labspace, and the greater fidelity to what they will see with physical machines the better. In addition, being able to easily simulate various catastrophic network incidents would be vital. So all in all, MAAS’s decreased ability (see above for the references to older blogs which had workable solutions) to handle a virtualbox powertype is seriously unfortunate.

It almost certainly wasn’t what motivated management’s reorganization. However, it does mean that I’m unlikely to spend any time testing workarounds at this juncture.