juju terminate-machine with local provider doesn't destroy machine

it would be useful to have the provider logs for the terminate-machine case, and --debug output from destroy-environment.

jamespage@armstrong:/etc/apparmor.d$ sudo juju --debug destroy-environment
2013-09-18 14:11:27 INFO juju.provider.local environprovider.go:32 opening environment "laptop"
2013-09-18 14:11:27 DEBUG juju.provider.local environ.go:210 found "10.0.3.1" as address for "lxcbr0"
2013-09-18 14:11:27 DEBUG juju.provider.local environ.go:234 checking 10.0.3.1:8040 to see if machine agent running storage listener
2013-09-18 14:11:27 DEBUG juju.provider.local environ.go:250 yes, don't start local storage listeners
WARNING: this command will destroy the "laptop" environment (type: local)
This includes all machines, services, data and other resources.

Continue [y/N]? y
2013-09-18 14:11:38 ERROR juju.container.lxc lxc.go:204 failed to destroy lxc container: error executing "lxc-stop": jamespage-laptop-machine-2 is not running
2013-09-18 14:11:38 ERROR juju supercommand.go:282 command failed: error executing "lxc-stop": jamespage-laptop-machine-2 is not running
error: error executing "lxc-stop": jamespage-laptop-machine-2 is not running

jamespage@armstrong:/etc/apparmor.d$ sudo lxc-list
WARNING: lxc-list is deprecated, please use lxc-ls --fancy.
         This symlink will be dropped in LXC 1.0.

NAME STATE IPV4 IPV6 AUTOSTART
----------------------------------------------------------------------------
jamespage-laptop-machine-1 RUNNING 10.0.3.20 - YES
jamespage-laptop-machine-2 STOPPED - - YES
jamespage-laptop-machine-3 RUNNING 10.0.3.11 - YES
jamespage-laptop-machine-4 RUNNING 10.0.3.95 - YES
jamespage-laptop-machine-5 RUNNING 10.0.3.107, 10.0.3.200 - YES
jamespage-laptop-machine-6 RUNNING 10.0.3.4 - YES
jamespage-laptop-machine-7 RUNNING 10.0.3.204 - YES
jamespage-laptop-machine-8 RUNNING 10.0.3.205 - YES
jamespage-laptop-machine-9 RUNNING 10.0.3.206 - YES

Thanks, James. lxc-stop is returning false because the machine is already stopped, and juju is (it seems) stopping there because of failure. We can either (a) have juju ignore lxc-stop failure, (b) just have juju run lxc-destroy -f which will implicitly stop the machine if needed (and not manually run lxc-stop), or (c) update lxc-stop to not return failure if the machine was already stopped.

I personally think both (b) and (c) seem reasonable, though someone may object to the lxc-stop change.

Certainly (b) will work around (c), so (b) by itself will fix this.