| 2015-08-22 19:11:26 |
Stéphane Graber |
description |
Hello,
While LXD 0.16 (our pre-FF release) does contain most features we were targeting for Ubuntu 15.10, we'd nevertheless appreciate being able to continue landing new upstream releases directly in Ubuntu until final freeze.
As it stands today, LXD is a universe package with a single reverse-dependency, nova-compute-lxd which is developed by the same team as LXD itself. We can and will commit to not breaking our REST-API (only supported interface) from now until release so any other piece of software using LXD wouldn't be affected by new releases.
Some of the things we have coming up which we'd appreciate being able to land this cycle still:
- Rework of our default bridge. We're currently using LXC's and that's not optimal as this bridge can cause subnet conflicts in some organizations.
- Resource limitation and reporting (cgroup limit configuration and querying)
- Security profile generation (on the fly apparmor & seccomp profile generation, improving container security)
As all our development and testing is on Ubuntu, we're very confident we can land new features in the archive without risking regressions.
The alternative to this would be to keep LXD in the archive at 0.16 and backport only critical bug fixes, keeping the rest of it mostly stale. The issue with this is that this would just encourage our users to use the daily PPA which we'd rather avoid as we'd like to have a good in-archive experience at release time. |
Hello,
While LXD 0.16 (our pre-FF release) does contain most features we were targeting for Ubuntu 15.10, we'd nevertheless appreciate being able to continue landing new upstream releases directly in Ubuntu until final freeze.
As it stands today, LXD is a universe package with a single reverse-dependency, nova-compute-lxd which is developed by the same team as LXD itself. We can and will commit to not breaking our REST-API (only supported interface) from now until release so any other piece of software using LXD wouldn't be affected by new releases.
Some of the things we have coming up which we'd appreciate being able to land this cycle still:
- Rework of our default bridge. We're currently using LXC's and that's not optimal as this bridge can cause subnet conflicts in some organizations.
- Resource limitation and reporting (cgroup limit configuration and querying)
- Security profile generation (on the fly apparmor & seccomp profile generation, improving container security)
- Improved LVM, btrfs and zfs storage support, including network operations (send/receive)
- Improved Ubuntu cloud image import and update
As all our development and testing is on Ubuntu, we're very confident we can land new features in the archive without risking regressions.
The alternative to this would be to keep LXD in the archive at 0.16 and backport only critical bug fixes, keeping the rest of it mostly stale. The issue with this is that this would just encourage our users to use the daily PPA which we'd rather avoid as we'd like to have a good in-archive experience at release time. |
|
