"Start Test Services" in test.yml is flaky

[trentm]

"Start Test Services" in test.yml is flaky. E.g.: https://github.com/open-telemetry/opentelemetry-js-contrib/actions/runs/26395884491/job/77696960078 (newlines added for readability)

Error response from daemon: failed to set up container networking:
  driver failed programming external connectivity on endpoint 
  opentelemetry-js-contrib-test-services-postgres-1 (538112839b818f127294ab3e985485f70df8f7e1a67e5c5951c8f96ca5b9b560): 
  failed to bind host port for 0.0.0.0:54320:172.18.0.3:5432/tcp: 
  address already in use

I've also seen flake failures related to pull docker images.

Perhaps this step could be retried?

[amang7788123]

Hi @trentm, I'd like to work on this, if you give a chance.
Looking at the failure log, it seems like two separate issues are surfacing under the same "flaky" symptom: one is a Docker network/port conflict (address already in use) likely caused by a leftover container from a previous run not being cleaned up properly, and the other could be a transient failure pulling Docker images during docker compose up.
My initial plan is to add a retry wrapper around the "Start Test Services" step using an action like nick-fields/retry, and also check whether adding an explicit docker compose down cleanup step before the test run would help prevent the port conflict in the first place.
Let me know if this aligns with what you had in mind, or if you'd prefer a different approach before I start working on a PR. Whatever it is please let me know.