HEALTHCHECK Instructions in Dockerfile

 Hello Guys in this post I wish to explore more intresting instructions HEALTHCHECK which can be used in Dockerfile. 

The most common requirement for any real-time projects monitoring using a side-car container which could run in parallel and try to check the process or application URL check or container reachability using ping command etc. 

Dockerfie instruction HEALTHCHECK

In Dockerfile we can have HEALTHCHECK instruction that allows us to know the condition of an application test runs as expected or not, when a container runs this will be returns a status as healthy, unhealthy based on the HEALTHCHECK command exit code. If the exit code 0 then returns healthy otherwise unhealthy.

HEALTHCHECK [options] CMD [health check command]

Example:

HEALTHCHECK --interval=3s CMD ping c1 172.17.0.2

here are the Healthcheck options 

1.  --interval=time in sec (duration 30s is default)
2.  --timeout=time in sec (duration 30s is default)
3.  --start-period=time in sec (duration 0s is default)
4.  --retries=3 () default 3 

Let's jump into experiment mode:

docker run -dt --name main-target busybox sh; docker ps 
docker inspect main-target

More specific

To get only IPaddress of a containter use the following format option:

alias dcip="docker inspect \
 -f '{{range.NetworkSettings.Networks}}{{.IPAddress}}{{end}}' "
ipdc main-target

Note: when you define the alias for ipdc important point don't miss space at the of the line.
Guys, here is glimpse of 'ping' command exit code:

ping -c1 google.com ; echo $?
ping -c1 shekhar.com ; echo $?

Observe that exit codes values, if a website exits returns 0 if not non-zero.

Get the main-target container IPAddress from the docker inspect command output. Now we will create a Dockerfile with the following code:

#File: Dockerfile
FROM busybox 
LABEL AUTHOR Pavan Deverakonda
HEALTHCHECK --interval=5s CMD ping -c 1 172.17.0.2

Note: You could find the IP from previous docker inspect command.

Let's call this image as monitor

docker build -t monitor .
docker run -dt --name monping monitor; docker ps

Observe the STATUS column for corresponding containers.

alias s="docker inspect -f '{{json .State}}' "
s monping |jq

docker inspect output filter with jq for HEALTHCHECK

USECASE 2: HEALTHCHECK for web applications

Let's see the 'curl' command usage in general

The following command returns HTML content which may be multiple lines

curl http://devopshunter.blog.com/test

Let's make this command usage in minimal way using -f or --fail options in curl command:

# Fail in silently single liner 
curl http://devopshunter.blog.com/test.txt -f

curl command with --fail or -f option

Run a container with a healthcheck command using. A Linux command that checks http uri using `curl` that returns HTML code or HTTP code as per the web applicaiton availability.

docker run -dt --name health-con \
 --health-cmd "curl --fail http://localhost" busybox 

Here we have not used any HEALTHCHECK options, so it will try to check by running health-cmd 30sec interval 3 retries and timeout for each as 30sec. that means after 2 minutes you can get the health status as 'unhealthy'. Because busybox don't run any web server inside the container.
We can check the health status of self container or other container which is accessable. Otherwise if it shared a network with other container which is running a web server.

#File: Dockerfile
FROM busybox 
LABEL AUTHOR Pavan Deverakonda
HEALTHCHECK --interval=5s CMD curl --fail http://localhost

Build the monitoring image that contains HEALTHCHECK instruction.

  
docker build -t moncurl .
docker images

For now we will test the same busybox container - unhealthy status.

docker run -dt --name moncurl-con moncurl sh
# Check the container health 
watch docker ps
  

#cleanup
docker rm -v -f health-con 

Now let's see the interval option how it will impact a container health:

docker run -dt --name health-con  --health-cmd "curl --fail http://localhost" --health-interval=3s busybox
watch docker ps 

My observation - at 0s(when container started) healthcheck starts after 3s test it, retries 3 times that means 3times 3s = 9s you will get the health status changed.

USECASE 3: HEALTHCHECK with Interval and Retries options

We can run a container to check the health with options interval and retries together as: 

 UNHEALTHY

docker run -dt --name health-con3 --health-cmd "curl -f http://localhost" --health-interval=3s --health-retries=1 busybox 
watch docker ps

HEALTHY

docker run -dt --name health-con3 --health-cmd "curl -f http://localhost" --health-interval=3s --health-retries=1 nginx 
watch docker ps

healthy status

Let's build a Healthcheck image

  
#File: Dockerfile
FROM nginx
LABEL AUTHOR Pavan Deverakonda
HEALTHCHECK --interval=5s --timeout=3s CMD curl --fail http://localhost || exit 1
EXPOSE 80

Now build the image

docker build -t moncurl:2 .
docker images

create the container with that image:

docker run -dt --name health-con2 moncurl:2 sh 

Please comment and share with your friends!