Docker HEALTHCHECK 在 nodejs express 应用程序中失败

Question

I have setup the new docker HEALTHCHECK command for a node / express container as follows

HEALTHCHECK --interval=5m --timeout=5s \
  CMD $ROOT_APPLICATION/healthcheck.sh localhost $PORT || exit 1

where the healthcheck.sh script do the following

function healthcheck () {
    pt=$2
    port=${pt:-"3000"}
    RES=$(curl -s http://$1:$2/status | jq '.status.online')
    if [ "$RES" == "true" ]; then
        return 0
    else
        return 1
    fi
}
healthcheck $@

The node api /status returns a json as follows

{
  "status": 
  {
     "online": true
  }
}

I'm sure that the health check is correct since within the container I can do

# $ROOT_APPLICATION/healthcheck.sh localhost $PORT
# status=$?
# [ $status -eq 0 ] && echo "success" || echo "failed"
success

Also calling manually the health check endpoint:

# curl -s http://localhost:3000/ws/1.0/status| jq '.status.online'
true

while if I check the container status I get a unhealthy state

CONTAINER ID        IMAGE                              COMMAND                  CREATED             STATUS                         PORTS                              NAMES
d8539fa50a40        my_container         "node --max_old_sp..."   35 minutes ago      Up 35 minutes (unhealthy)      0.0.0.0:3000-3001->3000-3001/tcp   my_image

[UPDATE]

Here is the output of docker container inspect

    "FinishedAt": "0001-01-01T00:00:00Z",
    "Health": {
        "Status": "unhealthy",
        "FailingStreak": 9,
        "Log": [
            {
                "Start": "2020-03-25T10:11:17.687207876Z",
                "End": "2020-03-25T10:11:17.829283718Z",
                "ExitCode": -1,
                "Output": "rpc error: code = 2 desc = oci runtime error: exec failed: container_linux.go:247: starting container process caused \"invalid environment 'NODE_ID'\"\n"

To reproduce this issue you must define docker-compose yaml file having the environment variable NODE_ID unassigned:

(minimal yaml file)

version: '2'
services:
 my_service:
     environment:
        NODE_ID
        NODE_ENV=production

Answer 1

Inspecting the service shows the results of the healthcheck (and a complete inspect would also show the issue with the NODE_ID variable).

starting container process caused \"invalid environment 'NODE_ID'\"\n

The unassigned NODE_ID in the compose file provided assumes you have defined this variable in the environment running the docker-compose up or docker stack deploy command. You can adjust this to give it a default value:

version: '2'
services:
 my_service:
     environment:
        NODE_ID=${NODE_ID:-undefined}
        NODE_ENV=production

For more details on that syntax, see https://docs.docker.com/compose/compose-file/#variable-substitution