[lxc-devel] LXC 2.0 (and related projects)

[Stéphane Graber]

Hello,

So you may have noticed we've been pushing new features to master for
months now and haven't done a new release yet.

That's mostly because I've been focusing so much on LXD and because LXC
basically just works so there wasn't much pressure.

I've now put out both LXC 1.1.5 and LXC 1.0.8 so stable users should be
covered for a little while.


Now with Ubuntu 16.04 LTS releasing in April 2016, we'd like to make
another long term support release of LXC, similar to what we did for
1.0. Except this time, we'd like to include lxcfs and lxd in there too.

Basically the plan is that by mid-February we'll have:
 - LXC 2.0.0
 - LXD 2.0.0
 - LXCFS 2.0.0

Now as for what this means for each project:

== LXC ==
LXC 2.0 will remain ABI compatible with LXC 1.0, so the ABI version will
just be bumped to 1.2 and not to 2.0 (so we'll still be liblxc1).

That's because nobody as came up with something which absolutely
requires breaking ABI, everything that was proposed so far can be done
by adding a new symbol and implementing a trivial backward compatibility
layer for the old symbol.

My plan is to tag the first LXC 2.0 beta in the next couple of weeks,
then tag more of those as changes land. Final release should be
mid-January.

== LXD ==
LXD will jump straight from 0.x to 2.0. With the first 2.0 beta being
tagged probably at some point in December.
We'll keep our fortnightly release schedule but tag betas and rcs
instead of stable release.

== LXCFS ==
LXCFS is going through some major rework right now, it was ported from
libnih to glib and from using cgmanager to doing straight cgroupfs.

It's now faster than ever before and no longer depends on cgmanager.

We'll give it a bit of time to stabilize but then will tag a 2.0 beta
for it too, making it jump from 0.x straight to 2.0 too.


== CGManager ==
What of CGManager?

Well, with the new lxcfs, it doesn't really need to exist anymore.
LXC and LXD can both be used using their regular cgroupfs backend over
lxcfs, so lxcfs alone is sufficient to provide container nesting and
unprivileged container management.

LXCFS is faster, more scalable and doesn't need any change to the client
software to interact with cgroups through it.

Dropping CGManager is therefore the most obvious option for us. We'll
keep maintaining it until it's no longer used by any supported LXC
release and will then kill the project entirely.






All that to say, it's time to look at those patches you've been stacking
on your side, dust them off and send them to lxc-devel if you want them
included in the next major LXC release.

It's also time to update your favorite templates, make sure they work
fine, pull the latest version of your favorite distro and generally are
in a shape you're happy with for the next 5 years.

I expect to all new features to land for at least another month, but the
sooner the better!

Thanks!

-- 
Stéphane Graber
Ubuntu developer
http://www.ubuntu.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.linuxcontainers.org/pipermail/lxc-devel/attachments/20151109/5e151333/attachment.sig>