Wait for Faros DB and Hasura/Metabase/n8n databases to exist before attempting to bring up other services

[ypc-faros]

Description

This change blocks Docker compose from attempting to bring up Hasura/Metabase/n8n until faros-init has created all the databases (Faros DB, Hasura/Metabase/n8n cfg databases). At the moment, we see a lot of failures/retries because these services attempt to connect to the database before it is available.

The issue has been reported here: https://faroscommunity.slack.com/archives/C0335HMN2NQ/p1648851029917679

Type of change

• New feature (non-breaking change which adds functionality)

Checklist

• Have you checked to there aren't other open Pull Requests for the same update/change?
• Have you added an explanation of what your changes do and why you'd like us to include them?
• Have you successfully run tests with your changes locally?

[tovbinm]

we have a wait-for script to do this in from the init container. is this not sufficient?

[tovbinm]

if you want to expand the check, then I think it should also be done from the container

[ypc-faros]

The check we have today in the init container is to wait for the postgres service to be ready before attempting to create the databases.

The change here is different. What I want to accomplish is for Hasura, Metabase and n8n to be started only after the databases have been created to avoid failures and restarts.

[tovbinm]

do you need to tweak any additional settings, e.g.

The options that can appear before CMD are:
    --interval=DURATION (default: 30s)
    --timeout=DURATION (default: 30s)
    --start-period=DURATION (default: 0s)
    --retries=N (default: 3)

https://docs.docker.com/engine/reference/builder/#healthcheck

[thomas-gerber]

@tovbinm @ypc-faros could this change (starting only after the databases have been created) also fix #45 ?

[ypc-faros]

do you need to tweak any additional settings, e.g.

The options that can appear before CMD are:
    --interval=DURATION (default: 30s)
    --timeout=DURATION (default: 30s)
    --start-period=DURATION (default: 0s)
    --retries=N (default: 3)

https://docs.docker.com/engine/reference/builder/#healthcheck

I used:

      interval: 10s
      timeout: 5s
      start_period: 1m
      retries: 5

Also checked that it doesn't wait the full start_period to begin checking. If the container becomes healthy according to the test before the start_period the dependent containers are started.

I also changed the test to check for the n8n database instead of the faros database, since the n8n database is the last created in db-init.sh

[tovbinm]

I would put something like this:

interval: 15s
timeout: 5s
start_period: 30s
retries: 5

[ypc-faros]

Done, thanks!