On 14/02/17 21:45, Adam Bolte wrote: > On 14/02/17 21:30, Marcin Kulisz wrote: >> On 2017-02-14 14:59:37, paul wrote: >>> There is a case for arguing that having dbus installed by default is not >>> a minimal setup and perhaps not in-line with user expectations, but IMO >>> broken functionality out of the box is also not in-line with user >>> expectations. >> >> IMO if it's broken functionality it's in Ansible (as I tryied to point out >> above) in Debian we're trying to be consistent across spectrum of our images. I did some more digging, and found this discussion: https://github.com/saltstack/salt/issues/23223 I haven't recreated a new AMI without dbus to test, but from the report (and the dpkg command on a local Jessie box) it looks like the timedatectl command is included in the systemd package, but doesn't actually work unless dbus is running. If I run timedatectl in a Jessie schroot without the dbus daemon running, I get the following: $ timedatectl Failed to create bus connection: No such file or directory $ I (as a Debian end user) would not expect timedatectl to be shipped in a broken state within an AMI, and obviously Ansible and Salt developers didn't expect that either. They probably both had a fall-back to another command when the timedatectl command did not exist. A quick scan over at bugs.debian.org suggests there are no bug reports about this against the systemd package, which I guess is a bit surprising given how long ago the Salt Stack issue on GitHub above was created. -Adam
Attachment:
signature.asc
Description: OpenPGP digital signature