[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Travis CI tests failing in master

Also I see this a lot:
1) ERROR: Failure: ProgrammingError ((_mysql_exceptions.ProgrammingError) (1146, "Table 'airflow.task_instance' doesn't exist") [SQL: u'DELETE FROM task_instance WHERE task_instance.dag_id = %s'] [parameters: ('unit_tests',)])

Sent with ProtonMail Secure Email.

‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
On Thursday, September 27, 2018 9:20 AM, Jarek Potiuk <Jarek.Potiuk@xxxxxxxxxxx> wrote:

> Hello Everyone,
> Seems that since yesterday the builds started to fail in Travis CI badly.
> And we need some urgent actions to fix it otherwise everyone developing
> Airflow is affected now.
> I am a bit fresh in Airflow so I am not sure how to handle those problems
> in an "emergency" way (and I am not sure if I am the best to propose
> solutions), so it would be great that more experienced people could help to
> solve it.
> It was not caused by any change - builds that worked before, started to
> fail now. The nature of those errors suggests transitive dependencies
> problems (they are all errors during installation of requirements).
> Currently the master of airflow-incubator is broken. Actually the problem
> has worsened - now we seem to have two distinct problems depending on
> python requirement.
> Error 1: Flask appbuilder
> Yesterday we had just this one (for both python 2.7 and 3.5):
> "flask-appbuilder 1.11.1 has requirement click==6.7, but you'll have click
> 7.0 which is incompatible."
> and resulting error:
> pkg_resources.DistributionNotFound: The 'click==6.7' distribution was not
> found and is required by flask-appbuilder
> You can see it failing in the master here:
> Likely the solution is to pin flask-appbuilder to one of the earlier
> versions or exclude 1.11.1 from valid versions, or limit flask-appbuilder
> to <1.11.1 (but I am not sure which one is the best solution). I will try
> some of that soon to see if that helps and let you know.
> *Error 2: pynacl build problem *
> This is only affecting python3.5. And started to appear only today in new
> pull requests (of several people including mine, so it's again some
> dependency problem).
> Exception: ERROR: The 'make' utility is missing from PATH
> -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
> Failed building wheel for pynacl
> Running clean for pynacl
> One of the failing builds is here:
> That one is tougher and seems to be related to
> This one is a bit tricky
> and there is no obvious solution - maybe we should limit pynacl to some
> earlier (which?) version ? Can anyone help with that ?
> J.
> -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
> Jarek Potiuk, Principal Software Engineer
> Mobile: +48 660 796 129