Update Socrata ETL GitHub Action

[mddilley]

Associated issues

Since this ETL queries the production VZ database, we need the production Airflow DAG to run production code only. Otherwise, a schema change in staging (like adding a new column, etc) could break the ETL. This happened when the GraphQL query to get crash data had in_austin_full_purpose in the filters in staging while the production database still uses austin_full_purpose.

See https://austininnovation.slack.com/archives/C013G4RNJ01/p1690432638099059?thread_ts=1690431847.716579&cid=C013G4RNJ01

Testing

URL to test:
n/a

Steps to test:

1. Give me a sanity check on the logic in this change please 🙏

I read the action as:

• Runs when code is pushed to master or production branch
• if there is a code change within atd-etl/socrata_export/**:
  • if the push was from production then update Docker image tagged with production
  • otherwise, update Docker image tagged with development (for Airflow local testing - I realize that I am breaking the convention we started and can change this back to latest if anyone feels strongly)

---

Ship list

• Code reviewed
• Product manager approved

[johnclary]

Looks good to me. I agree that using latest isn't ideal. Fwiw staging might be a more explicit tag name? In any case take my approval with a grain of salt since i don't do development here :)

[frankhereford]

Thanks Mike!

[mddilley]

@johnclary yep, i also am not sure about the tag. I picked development only because this DAG uses the environment variable to decided which image tag to run (development locally or production). Not sure how useful having the development tag with code from master is but we shall see.

[roseeichelmann]

yay thanks for doing this mike, looks good to me!

[johnclary]

matching the env var seems like a good enough reason to me 🙏

[patrickm02L]

Good to go 🚢