CDE deployment via caprover

[JessyBarrette]

CDE relies on docker-compose for deployments (see example). As a lot of the hakai infrastructure is migrating towards managing and handling applications deployments via Caprover. Some changes will be needed to correctely manage and deploy a CDE instance.

CDE relies on the following containers and associated builds/images:

• redis: "redis:alpine"
• db: "postgis/postgis:13-3.1-alpine"
• web-api: ./web-api/Dockerfile
• nginx: nginx:latest
• scheduler: ./download_scheduler/Dockerfile
• frontend: ./frontend/Dockerfile
• harvester: ./harvester/Dockerfile
• db-loader: same as harvester and use after the harvester.

To simplify the deployment, images of the different components will be maintained via this repository and will get build via a captain file like the following example

The docker-compose file will remain within the repo for local developments.

[steviewanders]

harvester: no container locally installed within a enviromment and run occasionally

db-loader: same as harvester and use after the harvester.

These will need separate Python containers no? Otherwise where are they running?

[JessyBarrette]

I'm thinking combining them together and uploading the db from each threads in the harvester to temporary tables which would then be ingested in a different stage via the already present db functions

[steviewanders]

Cool. So the harvester gets a Python container?

[JessyBarrette]

yes, but we'll have to think how it gets run.

• just once when triggered
• a cron job internally

[JessyBarrette]

4 containers are now made available via Hakai Institute packages