Docker HEALTHCHECK fails due to trying to access Martin at IPv6

[SmallhillCZ]

Hi, with the new version we got a problem, where HEALTHCHECK instruction (introduced in #1397) fails due to trying to access Martin at IPv6 (localhost is translated to [::1]), while Martin is listening on IPv4 in the container.

Tried to run the healthcheck command inside the running container and found out the following:

This fails, makes the container status unhealthy and prevents the container to be recognized by our reverse proxy (Traefik), thus not being online:

/ # wget --spider http://localhost:3000/health
Connecting to localhost:3000 ([::1]:3000)
wget: can't connect to remote host: Connection refused

This works okay:

/ # wget --spider http://127.0.0.1:3000/health
Connecting to 127.0.0.1:3000 (127.0.0.1:3000)
remote file exists

[CommanderStorm]

Good point.
Our default is to only bind to ipv4..

martin/martin/src/srv/config.rs

Line 6 in ba2e5d5

pub const LISTEN_ADDRESSES_DEFAULT: &str = "0.0.0.0:3000";

I don't know of better options than chaning it to this.
What do you think?

HEALTHCHECK CMD wget --spider http://127.0.0.1:3000/health || exit 1

[SmallhillCZ]

I don't know of better options than chaning it to this. What do you think?

Either that, or somehow rewrite the generated /etc/hosts at runtime and remove the localhost entry pointing to [::1], but just using 127.0.0.1 seems like a much simpler solution.