Thanks CLN. But I think it’s already marked in both README and setup.py that currently Airflow supports up to Python 3.7. So I think it’s already fine.
Thanks! XD On Thu, Oct 31, 2019 at 08:15 clhub...@gmail.com <clhub...@gmail.com> wrote: > Resolved: > Apache Airflow not supported on Python 3.8 > > I got it to work under Python 3.7 on MacOS Catalina - 10.15.1 (19B88) > Let me know if I should document this someplace for the community. I am a > former committer. > > Installation steps: > > - pip install apache-airflow > - You will see the following error during installation: > - ERROR: flask-appbuilder 1.13.1 has requirement > marshmallow<2.20,>=2.18.0, but you'll have marshmallow 3.2.1 which is > incompatible. > - pip install marshmallow==2.18.0 > - pip install 'apache-airflow[crypto]' > - Optional - I am installing it because I am switching from SQLite to > Postgresql. Want to install passwords encrypted. > - airflow initdb > - airflow webserver > - airflow scheduler > > > Regards, > CLN > > On Wed, Oct 30, 2019 at 4:41 PM clhub...@gmail.com <clhub...@gmail.com> > wrote: > > > Airflow Dev, > > MacOS 10.15.1/ apache-airflow 1.10.6 - retrieved the latest via pip. > > > > > > Airflow starts, I can view the console in the web browser and everything. > > I've got airflow.cfg configured to use Postgresql > > And I've successfully loaded my own test dags into airflow. > > > > The problem is the scheduler which throws the exception > > "ERROR - Exception when executing execute_helper" > > The bottom of the traceback says > > AttributeError: Can't pickle local object > > 'SchedulerJob._execute.<locals>.processor_factory' > > > > I dug a bit deeper into the Trace. > > > > Looking at job/scheduler.py the processor_factory function.... you can > see > > clearly where the exception when attempting to execute > > self._execute_helper() in a try block. > > > > self._execute_helper() executes utils/helper.py reap_process_group. > > I am assuming it's all downhill here once the reap process is executed. > > > > Any help here would be appreciated. > > I am really looking forward to getting AirFlow up and running. > > >
