canonical-ci-engineering team mailing list archive

[Andy Doan <andy.doan@xxxxxxxxxxxxx>]

Public apology - I confused the dates for this. I have questions for 
something next week:

* developing/updating hooks
The most efficient process I have is a joke. I add a "raise 
RuntimeError" to the end of the hook I want to work on. I then run my 
juju-deployer. This causes the step I'm interested in to fail. I log 
into the box and edit the hook on that system (keeping the hack for the 
exception). I then run "juju resolved --retry <unit>" over and over 
until the hook works how I want.

* developing reusable charms
It seems like a lot of charms become specific to an application. They 
wind up using things like tarballs for install/upgrade payloads. We are 
developing a lot of similar services, so how do we avoid having to write 
a charm for each service? The way I'm currently doing this is by 
allowing our base charms to take a bzr branch/revno as config 
parameters. As I understand it, this is unacceptable in prodstack.