Fix Docker HealthCheck unhealthy status #374
Conversation
| test: [ "CMD", "curl", "-f", "http://localhost:80/" ] | ||
| interval: 2s | ||
| timeout: 5s | ||
| retries: 10 |
There was a problem hiding this comment.
The EDC UI health check was reported broken by people who deployed it with docker swarm. They needed to manually disable it in the image in the end.
Thus we should double check the HEALTHCHECK statement in the Dockerfile. To actually test the Dockerfile's HEALTHCHECK directive, we might require either docker swarm or kubernetes to launch the container in a way, that its original healtcheck actually gets checked.
In Docker you can define a healthcheck in the Dockerfile with a HEALTHCHECK directive.
It is best practice to supply it and we do.
However, docker-compose files do not respect that original HEALTHCHECK directive, which causes our health checks to be untested.
There was a problem hiding this comment.
To actually test the Dockerfile's HEALTHCHECK directive, we might require either docker swarm or kubernetes to launch the container in a way, that its original healtcheck actually gets checked.
Unfortunately, I was unable to find conclusive documentation on this, but as far as I am aware, there is no way to make Kubernetes honor the HEALTHCHECK instruction.
Health checks are a Docker-specific feature; the underlying runtime, containerd, does not support them natively. Kubernetes does not use Docker, instead accessing containerd (or comparable runtimes) directly.
There was a problem hiding this comment.
So we could remove the HEALTHCHECK directive from the Dockerfiles but add them as examples to the docker-compose.yaml files.
|
Will have to be revisited after #437. Maybe it's already fixed |
|
I believe this PR is stale after #437 |
Related to #361 Fix EDC UI Dockerfile