How to view docker-compose healthcheck logs?

Question

Inside my docker-compose.yml, I have the following service healthcheck section. I want to know if MariaDB is actually ready to handle queries. A service named cmd is configured to depend on condition: service_healthy.

  db:
    image: mariadb:10
    environment:
      MYSQL_RANDOM_ROOT_PASSWORD: 1
      MYSQL_USER: user
      MYSQL_PASSWORD: password
      MYSQL_DATABASE: database
    healthcheck:
      test: ["CMD", "mysql", "--user=user", "--password=password", "--execute='SELECT 1'", "--host=127.0.0.1", "--port=3306"]
      interval: 1s
      retries: 30

This healthcheck does not work, shows that the service is unhealthy.

How do I check the output of the test CMD?

Sherry

Answer 1

You can use:

docker inspect --format "{{json .State.Health }}" <container name> | jq

Output:

{
    "Status": "unhealthy",
    "FailingStreak": 63,
    "Log": [
        {
            "Start": "2017-03-11T20:49:19.668895201+03:30",
            "End": "2017-03-11T20:49:19.735722044+03:30",
            "ExitCode": 1,
            "Output": "ERROR 1064 (42000) at line 1: You have an error in your SQL syntax; check the manual that corresponds to your MariaDB server version for the right syntax to use near ''SELECT 1'' at line 1\n"
        }
    ]
}

And look for the output section.

To get the Output only:

docker inspect --format "{{json .State.Health }}" mariadb_db_1 | jq '.Log[].Output'

For swarm mode use the folling format (thanks for pointing it out @shotgunner):

{{json.Spec.TaskTemplate.ContainerSpec.Healthcheck}}

Feel free to swap jq for whatever tool you use for json pretty print.

Vivisect

Hi, do you know any way to attach this logs to the same json-logfile that writes entrypoint? I'd like to see healthcheck logs executing docker logs, all of them in the same place – Frodi

In swarm mode and in manager {{json .State.Health }} not works. Use {{json .Spec.TaskTemplate.ContainerSpec.Healthcheck }} instead – Emma

If using jq anyway, all querying can be done through it; it's shorter and nicer: docker inspect $CONTAINER | jq '.[].State.Health'. – Affray

I wonder if the format of the first command is correct. It boasts a suspicious blank space. – Optometry

Answer 2

docker-compose ps will indicate the status of each service, including its health if healthcheck is defined. It's good for a basic overview.

% docker-compose ps
                Name                                Command                       State                                       Ports                            
----------------------------------------------------------------------------------------------------------------------------------------------------------------
remix-theme-editor_analytics_1            /bin/sh -c /analytics/run. ...   Up                                                                                   
remix-theme-editor_base_1                 /bin/bash                        Exit 0                                                                               
remix-theme-editor_flower_1               /entrypoint --environment  ...   Exit 137                                                                             
remix-theme-editor_frontend_1             /bin/sh -c perl -p -i -e ' ...   Exit 137                                                                             
remix-theme-editor_js-app_1               npm run                          Exit 0                                                                               
remix-theme-editor_mq_1                   docker-entrypoint.sh rabbi ...   Up (healthy)            15671/tcp, 15672/tcp, 25672/tcp, 4369/tcp, 5671/tcp, 5672/tcp
remix-theme-editor_mysql-migration_1      /entrypoint_mysql-migratio ...   Exit 0                                                                               
remix-theme-editor_mysql_1                /bin/sh -c /entrypoint_wra ...   Up (health: starting)   127.0.0.2:3308->3306/tcp                                     
remix-theme-editor_page-renderer_1        npm run start:watch              Up                                                                                   
remix-theme-editor_python-app_1           /entrypoint                      Exit 2                                                                               
remix-theme-editor_redis_1                docker-entrypoint.sh /bin/ ...   Up (health: starting)   6379/tcp                                                     
remix-theme-editor_scheduler_1            /entrypoint --environment  ...   Exit 137                                                                             
remix-theme-editor_socket_1               /entrypoint --environment  ...   Exit 1                                                                               
remix-theme-editor_static-builder_1       npm run watch                    Up                                                                                   
remix-theme-editor_static-http_1          nginx -g daemon off;             Up                      127.0.0.2:6544->443/tcp, 80/tcp                              
remix-theme-editor_web_1                  /entrypoint --environment  ...   Exit 1                                                                               
remix-theme-editor_worker_1               /entrypoint --environment  ...   Exit 1                                                                               
remix-theme-editor_worker_screenshots_1   /entrypoint --environment  ...   Exit 1     

If you want more details, use docker inspect in conjuction with docker ps -q <service-name>.

% docker inspect --format "{{json .State.Health }}" $(docker-compose ps -q mq) | jq
{
  "Status": "starting",
  "FailingStreak": 48,
  "Log": [
    {
      "Start": "2018-10-03T00:40:18.671527745-05:00",
      "End": "2018-10-03T00:40:18.71729051-05:00",
      "ExitCode": -1,
      "Output": "OCI runtime exec failed: exec failed: container_linux.go:348: starting container process caused \"exec: \\\"nc\\\": executable file not found in $PATH\": unknown"
    },
...

You can always debug the healthcheck yourself by simply executing your healthcheck code yourself. For example:

% docker exec -it $(docker-compose ps -q socket) nc -w2 127.0.0.1 5672
(UNKNOWN) [127.0.0.1] 5672 (?) : Connection refused

You can also do the same in shell:

% docker exec -it $(docker-compose ps -q socket) bash
root@b5da5207d344:~/src# nc -w2 127.0.0.1 5672
(UNKNOWN) [127.0.0.1] 5672 (?) : Connection refused
root@b5da5207d344:~/src# echo $?
1

Finally, you can simply use docker-compose up in the first terminal window, and docker-compose logs -f in another. This will display all logs from docker-compose-managed containers.

Karajan

Hi, do you know any way to attach this logs to the same json-logfile that writes entrypoint? I'd like to see healthcheck logs executing docker logs, all of them in the same place – Frodi

Answer 3

In Swarm Mode:

1. First use docker service ps service_name in the manager to find the failed task id and the corresponding node.

manager$ docker service ps service_name
ID                  NAME                 IMAGE            NODE                DESIRED STATE       CURRENT STATE               ERROR                              PORTS
liwww3qzg9dz        service_name.1       image_name:1.3   s-3                 Running             Running 27 seconds ago                                         
hcgxmwk2efj0         \_ service_name.1   image_name:1.3   s-3                 Shutdown            Failed about a minute ago   "task: non-zero exit (137): do…"  

In this example, hcgxmwk2efj0 is the task id and s-3 is the node name.

2. Then use docker inspect --format "{{json .Status.ContainerStatus.ContainerID }}" task_id in the manager to get the container id.

manager$ docker inspect --format "{{json .Status.ContainerStatus.ContainerID }}" hcgxmwk2efj0
"412b09d5244047b31471248fd9a0807e5ea42406fb8f5b1701df2244933e30c8"

3. Then ssh to that node, and use the command docker inspect --format "{{json .State.Health }}" container_id | jq to get the log of the healthcheck. (| jq is not necessary in this command)

s-3$ docker inspect --format "{{json .State.Health }}" 412b09d5244047b31471248fd9a0807e5ea42406fb8f5b1701df2244933e30c8 | jq
{
  "Status": "unhealthy",
  "FailingStreak": 3,
  "Log": [
    {
      "Start": "2021-09-07T06:10:05.233163051Z",
      "End": "2021-09-07T06:10:07.585487343Z",
      "ExitCode": 0,
      "Output": "... log 1 ..."
    },
    {
      "Start": "2021-09-07T06:10:37.644936244Z",
      "End": "2021-09-07T06:10:39.881196276Z",
      "ExitCode": 0,
      "Output": "... log 2 ..."
    },
    {
      "Start": "2021-09-07T06:11:10.16172012Z",
      "End": "2021-09-07T06:11:25.161912411Z",
      "ExitCode": -1,
      "Output": "Health check exceeded timeout (15s)"
    },
    {
      "Start": "2021-09-07T06:11:55.297395088Z",
      "End": "2021-09-07T06:12:10.302928565Z",
      "ExitCode": -1,
      "Output": "Health check exceeded timeout (15s)"
    },
    {
      "Start": "2021-09-07T06:12:40.371234778Z",
      "End": "2021-09-07T06:12:55.371393914Z",
      "ExitCode": -1,
      "Output": "Health check exceeded timeout (15s)"
    }
  ]
}

Uziel