Re: How to quickly re-deploy the same charm

2017-08-18 Thread Casey Marshall
With juju 2, you can also use --force-units to upgrade the charm even in an error state: juju upgrade-charm --path /path/to/mycharm --force-units mycharm The unit agent will then usually retry the hook that's failing automatically, but you can use `juju resolved` on the unit(s) without the

Re: How to quickly re-deploy the same charm

2017-08-18 Thread fengxia
Did `juju upgrade-charm`. Saw "Added charm...". It works, sometimes, not consistently. The new code changes are not reflected on juju machine. Why is that? Any way to force a reload? On 08/18/2017 09:40 AM, Tom Barber wrote: Yeah you need to do juju resolved charmunit/0 --no-retry to get it

Re: How to quickly re-deploy the same charm

2017-08-18 Thread Tom Barber
Yeah you need to do juju resolved charmunit/0 --no-retry to get it out of the error state so that the upgrade happens. On Fri, Aug 18, 2017 at 2:32 PM, fengxia wrote: > Hi Juju, > > So here is what I got: > > 1. juju status shows `mycharm` is in an error state. > > 2. Fixed

Re: How to quickly re-deploy the same charm

2017-08-18 Thread fengxia
Hi Juju, So here is what I got: 1. juju status shows `mycharm` is in an error state. 2. Fixed the bug and built a new version of the same application charm. 3. `juju upgrade-charm --path /path/to/mycharm mycharm It says `Added charm local:trusty/mycharm to the model". But nothing happens

Re: How to quickly re-deploy the same charm

2017-08-18 Thread Tom Barber
juju upgrade-charm is I suspect what you're looking for . Tom On Fri, Aug 18, 2017 at 1:58 PM, fengxia wrote: > Hi Juju, > > I'm testing a locally built charm using `localhost` (LXD) setup. Everytime > `juju deploy` will take a good 5-10 minutes just to download and install

How to quickly re-deploy the same charm

2017-08-18 Thread fengxia
Hi Juju, I'm testing a locally built charm using `localhost` (LXD) setup. Everytime `juju deploy` will take a good 5-10 minutes just to download and install python packages before the charm code runs. I'm wondering what's a good practice to cut down this in dev iteration? So if I build a