DevOps Classroom Series – 27/Jun/2020

Docker Storage Drivers

• Storage Drivers allows is to create data in the writable layer of the container
• We are already of Image Layers concept. For every RUN,ADD/COPy in the Dockerfile new layers are created.
• When we create a container, a Thin R/w Layer is available for each container to store the data in the container.
• Lets Experiment This with the following Dockerfile

FROM alpine
RUN mkdir /mycontent && touch /mycontent/1.txt

• Build the docker image and give name as layerdemo
• we would be using interactive mode to understand the behavior
• Alpine image has one layer and layerdemo will be adding a new layer. And every container created using layerdemo will add a new Thin R/W layer
• Run the following commands in the container 1

docker container run --name one -it layerdemo /bin/sh

• This container is dealing with 3 layers now but inside the container it will look as if it is one file system
• Internally the layers are combined and presented as one disk to container using storage drivers.
• Now lets try to make some changes

# Inside docker container one
cat /mycontent/1.txt
echo "hello" >> /mycontent/1.txt

• Since the layers are readonly whenever you modify any existing content, the file gets copied to read/layer and changes are recored in the R/W. This approach is called as copy-on-write

How to preserve the data changed by the docker container

• Persistence/Preserving the data even after the container is deleted can be acheived by storing the data outside container. To do this we have 3 options
  • volumes
  • bind mounts
  • tmpfs

Bind mounts

• are stored anywhere on the host system.
• Oldest option available for persistenting the changes
• Create a docker container with bind mount option

docker container run -it --name bindmount -v /home/ubuntu/test:/app alpine /bin/sh

df -h
touch /app/1.txt
exit
docker inspect bindmount
docker container rm bindmount

• Lets create a new container using the same storage as mount and the same data should be loaded in the new container

docker container run -it --name bindmount2 -v /home/ubuntu/test:/app alpine /bin/sh

• There is other way of using the same command

docker container run --mount type=bind,source=/home/ubuntu/test,target=/app <imagename>
docker container run --mount type=bind,source=/home/ubuntu/test,target=/app,readonly <imagename>

Volumes

• Created and managed by Docker.
• You create docker volumes and use that volume in the container.
• Docker Volumes can be create in the Dockerfile using instruction VOLUME

FROM openjdk:8
ARG url=https://referenceappkhaja.s3-us-west-2.amazonaws.com/spring-petclinic-2.2.0.BUILD-SNAPSHOT.jar
ENV myfilename=spring-petclinic-2.2.0.BUILD-SNAPSHOT.jar
RUN mkdir /app && cd /app && wget ${url}
VOLUME /app
WORKDIR /app
EXPOSE 8080
CMD ["java",  "-jar",  "spring-petclinic-2.2.0.BUILD-SNAPSHOT.jar"]

• Build the image
• Now execute following commands

docker volume --help
docker volume ls
docker container run -d spc:1.0
docker volume ls

• Lets inspect the volume

docker volume inspect <volname>

• Lets navigate to the mountpoint

cd <path>

• Now lets remove the container and look into volume

docker container rm -f voldemo
docker volume ls

• Volume is still available after container is deleted, so we dont loose data.
• Planning for volume creation in Dockerfile might not happen in all cases, the we need to create and mount volumes
• Volumes are of two types
  • anonymous volumes
  • named volumes
• Lets create a named volume

docker volume create myvol
docker volume ls
docker volume inspect myvol

• Lets create an alpine container with this volume mounted

docker container run -d --name alp1 --mount source=myvol,target=/test alpine sleep 1d
docker volume inspect myvol
docker container inspect alp1
docker container exec alp1 df -h
docker container exec alp1 touch /test/1.txt
docker container exec alp1 ls /test

• Now lets remove the container alp1 and create a new container alp2 with the same volume mount and see if the data is preserved

docker container rm -f alp1
docker volume ls
docker container run -d --name alp2 --mount source=myvol,target=/test alpine sleep 1d
docker container exec alp2 ls /test

• Lets try to share the data between two contianer using the same volume

docker container run -d --name alp3 --mount source=myvol,target=/test alpine sleep 1d
docker container exec alp3 touch /test/2.txt
docker container exec alp2 ls /test

• Docker volume has also an option of docker volume drivers which allows volumes to be created in nfs, aws s3 etc
• Now lets remove the containers

 docker container rm -f $(docker container ls -a -q)
 docker volume ls

• Volumes can be deleted by using rm

Tempfs mount

• Mounted on to tmpfs in memory
• Tempfs mount can be created using –tmpfs or –mount flag

docker container run -d --name tmp1 --tmpfs /app alpine sleep 1d
docker container inspect tmp1
docker container run -d --name tmp2 --mount type=tempfs,destination=/app alpine sleep 1d

Container communication in Single docker host

• Lets create two containers

docker container run -d --name cont1 alpine sleep 1d
docker container run -d --name cont2 alpine sleep 1d

• Every container gets an ip address lets see

docker container inspect cont1
docker container inspect cont2

• Now lets see if containers have network connectivy b/w then by executing

docker container exec cont1 ping cont2

• Now lets see if containers have network connectivy b/w then by executing

docker container exec cont1 ping -c 4 172.17.0.4

• We are able to ping from one container to other by ipaddresses but names are not resolving
• Refer Here for networking series
• Lets start networking in next class

Multi staged builds

• In CI/CD we build applications and for building applications if we want to use docker containers then this multistage build concept will be handy
• Let me give a simple example of a java application

git clone https://github.com/wakaleo/game-of-life
# java 8 installed and maven install
cd game-of-life 
mvn package

• In Traditional ways
  • Now after the Package is created, Ensure you have Dockerfile which creates the image
  • Create the image and push the image to docker registry
  • When you want to deploy application the pull from registry
• Multistage build
  • Build the package as part of your docker image build
  • Create a Dockerfile as shown

  FROM maven:3-openjdk-8 as builder
  RUN git clone https://github.com/wakaleo/game-of-life.git 
  RUN cd game-of-life && mvn package
  
  
  FROM tomcat:8
  COPY --from=builder /game-of-life/gameoflife-web/target/gameoflife.war /usr/local/tomcat/webapps/gameoflife.war
  EXPOSE 8080
  

• Now as part of building docker image application code gets build and a package is generated which will be copied into destination container

docker image build -t spcms:1.0 .

Docker logging

• docker logs command show the loggs by a running continer
• This docker logs will show the logs from STDOUT, STDERR
• In docker to configure logging, we have loggin driver plugins
• The supported logging drivers are over here