Hey,

I am working migrating some of our Dataflow pipelines from beam 2.4.0 and
facing issues with `Failed to install packages: failed to install workflow:
exit status 1`

My approach to debug this so far as has been building the package tar in a
Docker container and installing form there to see if there is any trouble
within my `setup.py`. I can build my package without issues in the Docker
container and running with `DirectRunner` also work but the above error
will still persist still when running with the `DataflowRunner`. I have
tried installing and running with `DirectRunner` Dataflow Docker images in
my case: `dataflow.gcr.io/v1beta3/python:2.6.0`
<http://dataflow.gcr.io/v1beta3/python:2.6.0> and this still has the same
issue.

How can I debug this issue? I find the Dataflow Runner context is quite
opaque to work with the Dataflow docs sadly don't have anything to help
with such issues either:
https://cloud.google.com/dataflow/docs/guides/troubleshooting-your-pipeline

Best wishes,
Chris

Reply via email to