Healthcheck for web container fails!

[Netzvamp]

There is an urgent error that crashes openproject docker compose deployments:

There is no curl anymore in the app container image, so the health checks fail and autoheal is restarting the web container, but they don't come up again.

web:
    ...
    labels:
      - autoheal=true
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8080${OPENPROJECT_RAILS__RELATIVE__URL__ROOT:-}/health_checks/default"]
      interval: 10s
      timeout: 3s
      retries: 3
      start_period: 30s

root@docker1:/docker/openproject# docker inspect openproject-web-1 --format '{{json .State.Health}}'
{"Status":"unhealthy","FailingStreak":79,"Log":[{"Start":"2026-03-11T15:36:43.524098518+01:00","End":"2026-03-11T15:36:43.60891546+01:00","ExitCode":-1,"Output":"OCI runtime exec failed: exec failed: unable to start container process: exec: \"curl\": executable file not found in $PATH"},{"Start":"2026-03-11T15:36:53.610644642+01:00","End":"2026-03-11T15:36:53.721577561+01:00","ExitCode":-1,"Output":"OCI runtime exec failed: exec failed: unable to start container process: exec: \"curl\": executable file not found in $PATH"},{"Start":"2026-03-11T15:37:03.723026179+01:00","End":"2026-03-11T15:37:03.847926613+01:00","ExitCode":-1,"Output":"OCI runtime exec failed: exec failed: unable to start container process: exec: \"curl\": executable file not found in $PATH"},{"Start":"2026-03-11T15:37:13.850434569+01:00","End":"2026-03-11T15:37:14.035559271+01:00","ExitCode":-1,"Output":"OCI runtime exec failed: exec failed: unable to start container process: exec: \"curl\": executable file not found in $PATH"},{"Start":"2026-03-11T15:37:24.036825754+01:00","End":"2026-03-11T15:37:24.19010835+01:00","ExitCode":-1,"Output":"OCI runtime exec failed: exec failed: unable to start container process: exec: \"curl\": executable file not found in $PATH"}]}
root@docker1:/docker/openproject# docker exec -it openproject-web-1 bash
app@openproject:/app$ curl
bash: curl: command not found

I've fixed it by commenting out the "autoheal=true", but we need curl in the container.

[CodeDok]

One possible workaround currently is using wget:

test: ["CMD-SHELL", "wget -q --spider http://localhost:8080${OPENPROJECT_RAILS__RELATIVE__URL__ROOT:-}/health_checks/default || exit 1"]

[bwildenhain]

No, that doesn't work, as the container contains neither curl nor wget.

[CodeDok]

Might depend on the tag you use. Im using the "full" version without slim and it works fine. I searched in /usr/bin/ for available options first:

[bwildenhain]

Okay, I am using slim, which has at least Perl (without LWP), so I used the following script to check instead (see attachment, also includes changes for postgres17)

docker-compose.override.yml.txt

    test: ["CMD", "perl", "-MIO::Socket::INET", "-e", "'$$s=IO::Socket::INET->new(PeerAddr=>\"localhost:8080\",Timeout=>5)or die;$$s->print(\"GET /health_checks/default HTTP/1.0\r\nHost: localhost\r\n\r\n\");$$r=join(\"\",<$$s>);exit($$r!~/PASSED/)'"]

[ukberry]

Similar to the approach provided by @bwildenhain I modified the test command to get this working on my local install. I chose ruby since the container is guaranteed to have ruby installed.

test: ["CMD", "ruby", "-e", "require 'net/http'; exit 1 unless Net::HTTP.get_response(URI('http://localhost:8080/health_checks/default')).is_a?(Net::HTTPSuccess)"]

I have provided a pull request to my changes if it helps progress a fix.

[DuaelFr]

It seems that the 17-slim images now have curl 🤷