feat(devservices): Add healthcheck wait condition and parallelize starting of containers

[hubertdeng123]

This starts containers in parallel and makes sure devservices waits until containers are healthy before succeeding.

[hubertdeng123]

I think relay sometimes doesn't pass the healthcheck, I may need to increase the retries there

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.85%. Comparing base (8aae9ef) to head (138b41f).
Report is 1 commits behind head on main.

✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #178      +/-   ##
==========================================
+ Coverage   92.56%   92.85%   +0.29%     
==========================================
  Files          22       22              
  Lines        1412     1470      +58     
==========================================
+ Hits         1307     1365      +58     
  Misses        105      105              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[joshuarli]

imo --wait should be default

[joshuarli]

can we also define a redis healthcheck in sentry's devservices config? otherwise we could run into Redis is loading the dataset in memory during migrations in sentry sync.py

[hubertdeng123]

imo --wait should be default

I agree

can we also define a redis healthcheck in sentry's devservices config? otherwise we could run into Redis is loading the dataset in memory during migrations in sentry sync.py

It is already defined:

https://github.com/getsentry/sentry-shared-redis/blob/9c15e58b7b254e2fd4621fb47d790c22d1eeaf3e/devservices/config.yml#L14

[hubertdeng123]

Decided to use in-house logic for healtchecks because it seems like --wait is not reliable for relay unfortunately. Healthcheck is failing when the container itself clearly is displaying as healthy. Spent a few hours trying to figure out the problem with docker compose, and decided this was better as in the future we can add logging to indicate specific services have passed these healthchecks.

[hubertdeng123]

Docker container names are unique so this will be safe. Decided to make this change so that we could print a better status message if containers fail healthchecks

[IanWoodard]

Looks good! Left a few comments but otherwise we should be good to go

[IanWoodard]

Nit: since we don't use container_names anywhere else, why don't we combine these two lines?

[hubertdeng123]

I actually opted to do this here, since I think it's more readable than

containers_to_check.extend(get_container_names_for_project(cmd.project_name, cmd.config_path))