| Commit message (Collapse) | Author | Lines |
|
I've restructured the build & deploy sequence in our GitHub Actions CI
workflow:
- There's now a separate workflow to deploy to our cluster
- The Build workflow has an additional `if` check to ensure we're only
starting the build & deploy sequences for pushes to master
- We now checkout the python-discord/kubernetes repo to fetch the
deployment.yaml file located in there
|
|
|
|
|
|
The CI was failing on occasion because postgres did not have time enough to start properly before we started running our makemigrations check. I've moved the docker-compose step to earlier in the workflow to give it some time to warm up.
|
|
|
|
I've also restructured it a bit by using reference links instead of
inline links.
|
|
Unfortunately, the way we previously set up our workflow caused flake8
to run on code already committed to master, not the changes made in a
PR, because it ran in the context of the target branch.
This is obviously useless when it comes to protecting our codebase from
linting errors. I've now set up flake8 in a different way, using
Workflow Commands to create error annotions.
I've also split up the workflow into two separate workflows.
|
|
|
|
|
|
|
|
|