Healthcheck in docker compose

[tbfpartner]

I am trying to implement a docker compose health check but it fails.
I cant find any documentation on what endpoint i need to use.

    # healthcheck:
    #   test: curl --fail --silent -I http://127.0.0.1:3000 || exit 1
    #   interval: 60s
    #   timeout: 5s
    #   retries: 5
    #   start_period: 180s

i guess i need a specific endpoint like /health or /api/v1/health?

[C4illin]

curl -f http://localhost:3000 should work, maybe it would be good to add a specific endpoint though

[tbfpartner]

thanks... that sorta works... any ideas why it returns a http-500 when using the -I flag?

[tbfpartner]

actually just asked my local LLM :) maybe this helps for a future feature... lets hope PR #375 goes live some day. it looked promising.

Okay, you're facing a common issue with health checks and servers that sometimes return 500 errors even when "technically" functioning. The -I flag (HEAD request) is much more sensitive to errors than a full GET request. Your server seems to respond with content on a full request but fails on the HEAD request, which is likely causing your healthcheck to fail.

Here's a breakdown of why this is happening and how to adjust your Docker Compose healthcheck, with several options, ranging from simplest to more robust:

Understanding the Problem

HEAD vs. GET: A HEAD request only asks for the headers. It's a quick way to check if the server is reachable and responsive without downloading the entire page. Your server is handling the full request (curl --silent http://localhost:3000) but failing the minimal request of HEAD. This could be due to a temporary issue initializing something before the headers are sent, or a dependency not being quite ready.
Docker Compose Healthcheck Purpose: Docker Compose healthchecks are designed to determine if a container is "healthy" before it's considered part of a service. This is important for orchestration – Docker won't try to route traffic to unhealthy containers.
--fail in curl: The --fail option makes curl exit with a non-zero status code if the HTTP request fails (status code >= 400). This is precisely what's causing your healthcheck to fail.

[C4illin]

Sound like a reasonable answer, but I dont have any idea why it might fail

[C4illin]

I added /healthcheck which returns a simple json