Using a custom ROCKET_ADDRESS value results in erroneously reporting container as "Unhealthy" (healthcheck.sh)

[ghost]

Subject of the issue

healthcheck.sh ignores ROCKET_HOST value (if different from "localhost") and erroneously reports container as Unhealthy

Deployment environment

Docker installation of vaultwarden with a little customization (set ROCKET_HOST and ROCKET_PORT in the environment section of docker-compose).

• vaultwarden version:
  vaultwarden:latest

• Install method:
  docker / docker-compose

Steps to reproduce

Set ROCKET_HOST in docker-compose environment section for the container to a real local IP address (e.g. 192.168.1.1).
Set "network_mode: host" in docker-compose.
Start up the container.
After a few minutes, check container status with "docker ps".

Expected behaviour

"docker ps" reports container as "Healthy" if healthy, regardless of whether ROCKET_HOST is "localhost" or a different value.

Actual behaviour

"docker ps" reports container as "Unhealthy" if ROCKET_HOST value is different from "localhost".

Troubleshooting data

Issue stems from the very last line of /healthcheck.sh, which hardcodes "localhost" as the destination host for the curl call.

Is:

curl --insecure --fail --silent --show-error \
     "http${s}://localhost:${ROCKET_PORT}${base_path}/alive" || exit 1

Should be (with the proper initialization of ROCKET_HOST earlier, of course):

curl --insecure --fail --silent --show-error \
     "http${s}://${ROCKET_HOST}:${ROCKET_PORT}${base_path}/alive" || exit 1

Consequently, if Rocket doesn't listen on localhost:$ROCKET_PORT, but on (e.g.) 192.168.1.1:$ROCKET_PORT, the healthcheck will fail and the container will be reported as Unhealthy.

[jjlin]

I assume you meant ROCKET_ADDRESS, not ROCKET_HOST? The latter isn't a thing, AFAIK. Anyway, #2844 should fix this.

[ghost]

Correct, ROCKET_ADDRESS. Tired when writing that. Love the super-quick fix.