[19:17] <CtfINC> Please look at Bug #1436956, Critical system files such as /sbin/shutdown and /sbin/init missing after updating upstart. Proposed importance: Critical
[19:23] <rbasak> CtfINC: we're using systemd by default in Vivid now. So I don't think the steps to reproduce are particularly clear here.
[19:23] <rbasak> Is this a non-default system configuration?
[19:24] <teward> rbasak: should that bug be marked "Vivid uses systemd by default, now, is this on a non-default configuration?" and status.set(Incomplete)?
[19:24] <rbasak> Binaries are shipped in upstart-bin now I believe.
[19:24]  * teward ran out of things to do nowadays
[19:24] <rbasak> If you like :)
[19:25] <teward> done because i'm bored :)
[19:25] <rbasak> Thanks.
[19:25] <rbasak> armhf suggests it to me. I want to know how the system was installed.
[19:26] <teward> ooo ubuntu-server daily crashed my VMware guest enviro o.o
[19:26] <teward> that concerns me
[19:26] <teward> (sorry, i'm babbling, i'll be silent)
[19:26] <rbasak> It sounds like the system doesn't have a suitable metapackage pulling upstart in or something.
[19:29] <CtfINC> rbasak: Yes this is a system with both systemd and upstart installed. It is the default Docker image provided by labs.online.net
[19:30] <teward> *why* do they even offer a 15.04 image
[19:30] <teward> they're stupid to do so, just saying
[19:31] <rbasak> CtfINC: probably best to start by reporting the bug to them.
[19:32] <rbasak> CtfINC: I believe the binaries are provided by upstart-bin now. So the question is why it hasn't been pulled in by a suitable metapackage on upgrade (or similar).
[19:32] <CtfINC> rbasak: Yes I plan to also report to them. The image is under "InstantApps" and named "Docker 1.3.2 + fig + nsenter + pipework + gosu". Currently they require an invitation to sign up, but may be launching soon.
[19:32] <rbasak> CtfINC: it could well be an installer bug, but presumably the Docker image wasn't constructed by an installer shipped by Ubuntu.
[19:33] <rbasak> CtfINC: please could you update the bug with this information?
[19:33] <CtfINC> rbasak: I looked at the changelog of upstart package, the problem is that those binaries are moved from upstart to upstart-sysv (a new package), thus missing in upstart latest version.
[19:34] <rbasak> Ah, right.
[19:35] <CtfINC> rbasak: And the system does not require upstart-sysv to be installed during the "apt-get upgrade" process. Hence the binaries are missing and system is dead.
[19:35] <rbasak> CtfINC: though now in Vivid systemd is default.
[19:35] <rbasak> CtfINC: so anybody wanting to use upstart needs to install the correct package manually.
[19:35] <CtfINC> rbasak: Thanks I will open a ticket with online.net
[19:35] <rbasak> CtfINC: OTOH, the bug might be that it didn't get upgraded from upstart correctly.
[19:35] <rbasak> (to systemd)
[19:36] <rbasak> But I believe that has worked correctly for everybody else, so it's likely something missing in their image build process.
[19:36] <CtfINC> rbasak: If you need further information I'd be happy to provide.
[19:37] <rbasak> CtfINC: I'd say this bug is very close to "Invalid" for Ubuntu right now - no known way to reproduce on an Ubuntu system. Maybe I don't have all the information though, but in the first instance it needs to be addressed by the vendor of your Docker image.
[19:37] <rbasak> I'm not sure what else I can ask, apart from a technical analysis from them on why this is a bug in Ubuntu as opposed to their image build process.
[19:37] <CtfINC> rbasak: By the way I just found another bug report a couple of days ago, with the same issue
[19:38] <CtfINC> rbasak: Please see Bug #1436691
[19:39] <rbasak> CtfINC: it's still seems unlikely to me. That reporter has provided no information on system details.
[19:40] <rbasak> CtfINC: if there were really a "system can't boot" on upgrade within Vivid bug, I'd expect dozens of reports within a day.
[19:40] <CtfINC> rbasak: The latest version of upstart package moves the binaries to upstart-sysv, but it does not depend on upstart-sysv. I think the problem is here?
[19:40] <rbasak> I'm open to finding some kind of weird edge case bug of course.
[19:40] <rbasak> CtfINC: but all Vivid systems should successfully have switched to systemd already.
[19:40] <rbasak> Maybe there's a bug somewhere that causes that to fail.
[19:41] <CtfINC> rbasak: Thanks for the responses