[xen-tools-dev] Feature freeze for Xen-Tools 4.2 (and Debian Squeeze)

[Dmitry Nedospasov]

Hey Devs,

On Fri, Aug 06, 2010 at 04:30:39PM +0200, Axel Beckert wrote:
> I just merged in (well, cherry picked) Stéphane's btrfs support and
> created a new branch "4.2".
> 
> That branch should only get bugfixes, no more new features and the
> final 4.2 (and possibly subsequent 4.2.1 or maybe also 4.3) will be
> release from there. If you are not sure if a change should be also
> applied in that branch please ask before pushing.

Perfect, this is the best way to go.

> All new feature should go into the master branch (heading towards
> 5.0), bugfixes in both branches. Looking forward to bigger code
> refactorings, especially those which reduce the tons of code
> duplication.

Well, for 5.0 I have a couple of ideas and fixes that we devs should
discuss, I especially want to get rid of the code duplicaton ASAP.
This would fix several issues, I'll name just one example that comes
to mind:

Since Debian doesn't have a per distribution hook folder, xen-tools
always installs a -xen kernel. This isn't necessary starting with
squeeze as everything has moved to ParaVirtOps (which the standard
kernel supports, though only as domU obviously).

I think we should start a branch "common", and slowly migrate to more
common scripts. I'm really tired of implementing every fix an N
number of times (where N is the number of distros we support).

Let's get this painful step over with! Who's with me?



In terms of 4.2, I might look into doing a couple of bugfixes. theres
some really wierd exit codes that go on in xen-tools, that i think
should still be fixed in 4.2. Try this one:

Look what happens when you run xen-create-image either when your
networking or DNS resolution is down (i.e. no nameserver in
resolv.conf). The Error is completely undescriptive, so we need to set a
different FAIL value somewhere. The part i like the most, is that the
last error you see is that the ssh keys couldn't be found :)

We fixed debootstrap not being installed right? Because I can remember
the output being very similar. In fact i think any other type of
debootstrap fail, also generates such an incorrect output. Try
specifying a mirror that doesn't exist, IIRC - same thing.



On a slightly different Note we need to figure out a way to quickly
report bugs. I'm seriously contemplating just writing this list every
time i find a bug, but it seems like there might a better way to do it,
that doesn't seem as annoying. Are there any lightweight bugtrackers we
can use? Submitting every bug to debian, also seems like a bit much.

> Additionally I added Stéphane Jourdois aka kwisatz to the list of
> xen-tools developers on gitorious.org. He should be able to push to
> the official git repository from now on, too. Welcome and looking
> forward to your further contributions, Stéphane!

Sweet! Welcome, Stéphane!

> In other news: Debian 6.0 Squeeze is frozen[2], nearly no new features
> will move from unstable to testing. I though hope to get the final 4.2
> into Squeeze.

So what does this mean for us? Although I know am back to running Debian
dom0s, I still run the latest gitorious version in general :)

@Axel: How much time do we have for some final Bugfixes? I've been
really busy these past couple of weeks :(

That reminds me, the xen guys still haven't fixed the pygrub bug, which
i found [1]. Ironically it's been like a stepping on rakes :) I even
complained to Axel one evening when upgrading a server, and it turns out
that it was yet again that issue :) Happened to me again today, after I
upgraded my xen-hypervisor package (my local changes to the python
script got overwritten).

Regards,

D.

[1] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=588811
-- 
Dmitry Nedospasov <dmitry at nedos.net> -- Twitter: @nedos
Web: http://nedos.net -- Github: http://github.com/nedos