Draft: Resolve "Refactoring the infrastructure"
Closes #26
Update workflow:
-
using only a single staging
branch with multipledev
merge requests -
the main deployment happens based on versioned tags (support rollback when needed) -
protect main
andstaging
branches -
plan how to test notebooks individually (binderhub) -
plan how to update major packages (hackathons) -
plan how to run notebooks anonymously (and safely) -
make sure that "short-urls" still works -
make sure that sensitive notebooks have not been shared publicly (Please note that analytics
is a public repository)
Update Dockerfile:
-
Multi-stage build (builder, base, final) -
predetermine common package requirements -
cron job to rebuild image regularly -
fine-tune header generation (using as a preprocess instead of running it at build time) -
revisit example data storage (right now it is 3.1G) (#18)
Update CI/CD pipeline:
-
update jupyterhub's chart -
To get rid of ananlytics-deployment
dependencies.-
Store all sensitive data in env variables -
Move all deployment configurations here
-
-
having cleanup processes (images, environments) -
using ingress controller