fix: modified container healthcheck to support custom port via -port/--port flag

[dylandacosta8]

Problem Statement

The Dockerfile currently has a static healthcheck URL (http://localhost:8080/healthz) that doesn't adapt when users specify a different port using the -port flag. This causes healthchecks to fail when running cAdvisor on non-default ports.

Related Issues

Fixes #3618

Root Cause

Environment variables set in the entrypoint script don't persist for HEALTHCHECK commands because they run in separate process contexts. Docker doesn't share the entrypoint's environment with the healthcheck process.

Proposed Solution

Implemented a dedicated healthcheck script that:

1. Reads the cadvisor process command line from /proc/1/cmdline
2. Parses the -port flag to extract the actual port
3. Performs the healthcheck on the correct port
4. Falls back to 8080 if no custom port is specified

This approach works because Docker healthchecks run inside the container and can access the process information.

Breaking Changes

None - fully backward compatible

Test Case 1: Default Port (8080)

docker build -f deploy/Dockerfile -t cadvisor-test .
docker run -d --name cadvisor-default cadvisor-test
sleep 35
docker inspect cadvisor-custom --format='{{json .State.Health.Status}}' # Should show "healthy"

Test Case 2: Custom Port (9090)

docker run -d --name cadvisor-custom -p 9090:9090 cadvisor-test -port=9090
sleep 35
docker inspect cadvisor-custom --format='{{json .State.Health.Status}}' # Should show "healthy"

Test Case 3: Verify Healthcheck Logic

docker exec cadvisor-custom cat /proc/1/cmdline | tr '\0' '\n'  # Should show -port=9090
docker exec cadvisor-custom ./usr/bin/healthcheck.sh  #Should exit with status 1

Note:

Individual CLA submitted

[dylandacosta8]

Added the required boilerplate to the shell scripts

[dylandacosta8]

Hey, could a maintainer help me understand if any changes are required here?