Healthcheck for web container fails!

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.

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"]

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

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:

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/)'"]

@bwildenhain

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.

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

Navigation Menu

Repository navigation

Healthcheck for web container fails! #150

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Healthcheck for web container fails! #150

Description

Activity

CodeDok commented on Mar 11, 2026

bwildenhain commented on Mar 12, 2026

CodeDok commented on Mar 12, 2026

bwildenhain commented on Mar 12, 2026

ukberry commented on Mar 12, 2026

DuaelFr commented on Aug 5, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions