Chapter6.md

Chapter 6 - Persistence

Containers Are Designed To Be Ephemeral

Although we can start stopped containers, and pick up where we left off, that's an anti-pattern. Ideally we want repeatable operations on repeatable infrastructure.

Most worthwhile things are performing operations on data, and there are 3 ways to handle data volumes on containers.

Every time you run docker run it's a brand new container based on the image

The Container Lifecycle

Looking back at our hello-world container from earlier, let's run it again:

docker run --name myhelloworld hello-world

Now if we check the list of containers:

docker container ps You'll notice it isn't there.

But if we get the list of all containers:

docker container ps -a

We can now see it:

CONTAINER ID        IMAGE               COMMAND                  CREATED             STATUS                       PORTS               NAMES
f9850a139a45        hello-world         "/hello"                 32 seconds ago      Exited (0) 31 seconds ago                        myhelloworld

We can see that it Exited(0)

This means the container is 'Stopped', stopped containers still exists on your disk, but are not using any other system resources. By default they will persist until you remove them.

Now compare that with our alpine containers:

docker run --name myalpine -dit alpine

Checking to see running containers:

docker container ps

We can see it's still running:

CONTAINER ID        IMAGE               COMMAND             CREATED             STATUS              PORTS               NAMES
a71762793e40        alpine              "/bin/sh"           5 seconds ago       Up 4 seconds                            myalpine

What's the difference? Write down your ideas of why these two containers appear to act differently.

Naming Containers

In the last section we started containers specifying their names, myhelloworld and myalpine. What happens if we try to start a container with the same name?

docker run --name myalpine -dit alpine

We get an error.

Well it makes sense we can't have two running containers of the same name. So let's stop the running container:

docker container stop myalpine

Now if we try to run again:

docker run --name myalpine -dit alpine

We still get an error.

The rule is we can't overwrite a container, if you read the error message it says we need to remove the container to be able to reuse the name. So let's do that:

docker container rm myalpine

What scenarios can you think of where this might be a problem? Can you find in docker run --help a way of managing this?