MAAS + LXC containers == unable to upgrade charms due to git hostname resolution issues

[James Page]

I know that git has disappeared from use by juju on service units for 1.19.x, but I still see issues upgrading charms running in LXC containers in MAAS environments under 1.18.x; specifically I get:

2014-06-12 10:15:26 ERROR juju.worker.uniter.charm git.go:211 git command failed: exit status 128
path: /var/lib/juju/agents/unit-nova-cloud-controller-0/state/deployer/update-20140612-101526680418371
args: []string{"commit", "--allow-empty", "-m", "Imported charm \"local:trusty/nova-cloud-controller-315\"."}

*** Please tell me who you are.

Run

  git config --global user.email "<email address hidden>"
  git config --global user.name "Your Name"

to set your account's default identity.
Omit --global to set the identity only in this repository.

fatal: unable to auto-detect email address (got 'root@juju-machine-0-lxc-6.(none)')

2014-06-12 10:15:26 ERROR juju.worker.uniter.filter filter.go:116 tomb: dying

[James Page (james-page)]

I worked around this by adding an entry to /etc/hosts for 127.0.1.1 so that git was able to resolve a full hostname.

[Curtis Hovey (sinzui)]

Juju 1.20.0 doesn't use git, so there is no plans to fix this bug in 1.18.

[James Page (james-page)]

Bearing in mind that 1.18 is the current stable release shipping in 14.04 and its taking a while for 1.20 to stablize in full, please can this bug be re-assessed for inclusion in 1.18.4.

It's not a great feeling watching your message broker being restarted continually in an 1.18.4 deployed OpenStack when you try to upgrade the charm.

For anyone hitting this, setting a fully qualified hostname in /etc/hostname and /etc/hosts resolves the problem.

[Kapil Thangavelu (hazmat)]

perhaps configure the info git is requesting to avoid the implicit fqdn host lookup failure, ie. juju run --all git config --global user.email "<email address hidden>"

[Anastasia (anastasia-macmood)]

targeted against old milestone