Hi Ben, I've run into a similar issue and so far have a sub-optimal solution. The main issue is that master is unstable, at least for my setup. I filed an issue here: https://issues.apache.org/jira/browse/AIRFLOW-664 So in the meantime, I've made a branch from 1.7.1.3 and apply my changes there, pip installing from the github commit. I'm also making pull requests, but since I can't run master, I'm not able to test or feel confident I'm not introducing a regression. This problem with master might be with my particular configuration -- at least I haven't heard anyone else having this problem, and I've asked on gitter, the dev mailing list, and filed a Jira issue.
For upgrading, I have a deploy process that calls `airflow initdb` on each deploy. `initdb` calls `upgradedb` so In the case of bumping the version, it handles the schema upgrade. I'm not sure if there is a graceful way to handle downgrading, but on the few times I've had to do that, I've either restored MySQL from a snapshot or just `airflow resetdb` depending on the situation. cheers, Dennis On Thu, Dec 1, 2016 at 10:54 AM Ben Hoyt <benh...@gmail.com> wrote: > I'd like to start using the EMR hooks and operators (for example > > https://github.com/apache/incubator-airflow/blob/master/airflow/contrib/hooks/emr_hook.py > ). > However, they were added on June 30, and the latest release is 1.7.1.3 > dated June 13. So a couple of questions: > > 1) What is the best practice if I want newer features? Is running off the > latest master a terrible idea? How do other folks do this? > 2) What's the release schedule? From the commit history it looks like a lot > of stuff has been added in the last few months but isn't available in a > release yet. > > Somewhat relatedly, I didn't see any documentation on upgrading. I presume > there may be database migrations involved? How does one upgrade the db from > version X to version Y, or know if migrations are needed? > > Thanks, > Ben >