Comment 0 for bug 1427654

As per https://blueprints.launchpad.net/ubuntu/+spec/core-1411-systemd-migration we aimed for switching the system (not session) init from upstart to systemd this cycle. As this was by and large a 1.2 man show, this took a little longer. One of the remaining blockers for switching the default was to fix NFS (bug 1312976) which is now ready to be uploaded. There is also juju (bug 1409639, landed upstream now) and maas (bug 1423613, scheduled), but for those we can work around missing systemd support by installing upstart on instances.

We are a few days past feature freeze now, bug Steve and I discussed last week, and we still want to aim for switching the default in vivid. To be clear, this will affect Ubuntu desktop/server/cloud and the flavours like Kubuntu, but *NOT* ubuntu-touch. Migration to systemd is blocked on touch (too old kernels, some unported jobs), and was not scheduled for vivid. For that we need to fork ubuntu-minimal-touch to keep upstart (there is a work item on the spec).

I know from at least Kubuntu and Ubuntu GNOME that they are pushing for switching to systemd. We should get an ack from the others (Lubuntu/Xubuntu/MATE) before doing the switch; but quite a lot of vivid users are running systemd already, and I believe we shook out the worst bugs now.

So if/once the release team generally agrees, these are the things which need to happen before switching:

  * Signoff/test from remaining flavours (MATE, Xubuntu, Lubuntu)
  * Land NFS
  * Fork ubuntu-minimal-touch

Operationally, the switch is merely to change s/upstart/systemd-sysv/ in the seeds, i. e. flip the dependency in ubuntu-minimal. I did that in https://launchpad.net/~pitti/+archive/ubuntu/systemd, fixed ureadahead, and verified that upgrades from trusty and utopic cleanly upgrade to vivid+PPA with removing upstart and installing systemd-sysv.

Around that time I will send a mail to ubuntu-devel-announce@ to warn about the impending switch, and give some pointers how to boot with upstart (from grub menu or reinstalling upstart), how to debug boot/shutdown issues, etc.

Contingency plan: This will trigger wider testing, and most certainly uncover more integration issues. I'll try to keep up with them, but if we get to a point where we find that there are too many/too serious regressions, we can flip the seed back to upstart and get back to the situation today.