I agree, I'm definitely not on the side of supporting CentOS, Arch, or
whatever version of system VM an admin wants, it would be a nightmare for
the people who maintain the systemvm scripts. That's what I was attempting
to say. But I do think that it should be documented well enough that people
who want to roll their own Debian system, (because ours is out of date, or
has a slow driver, or whatever) can just install the scripts and run with
it.

Personally I'm more worried about the userspace/file structure differences
than library or kernel versions when it comes to making an OS work as a
system vm. Shell scripts that run iptables, wget, tar and ifconfig commands
tend to be very portable and distro agnostic, but differences in upstart,
systemd, sysvinit, chkconfig and analogs, config file locations are all a
pain for our system shell scripts. So it seems reasonable to have good
instructions on how the system vm is put together on Debian, and if someone
wants to drop in their own kernel, update the OS, or perhaps use Ubuntu
they can do that fairly painlessly. It will also facilitate someone
submitting a better system VM for consideration as the main one, or at
least help broaden the base of developers maintaining the default system
vm. Perhaps the existing documentation is sufficient for this, but I've
seen several questions about it.

In any case, it should be clear that people are on their own if they don't
use the provided image, or at the mercy of whatever third party is offering
system vms they choose to use. Perhaps that should be a part of the bug
form. I just don't want people to feel like they're being held hostage to
the one CloudStack offers because they can't find sufficient information.
After all, the one currently offered is almost a year old now, with no
updates, security or otherwise (maybe I'm wrong on this; it seems the
systemvm is the acton one from last Feb).

Reply via email to