configmap.md

ConfigMaps

From the k8s definition, "ConfigMaps allow you to decouple configuration artifacts from image content to keep containerized applications portable"

So what are they really? Basically, they're the configuration portion of a microservice where you swap in and out environment variables. Want to point to a specific database, namespace, etc.? You're probably going to be doing that with ConfigMaps, unless it's more sensitive data in which you'll be using secrets.

Here's a quick Node.JS example connecting to a local MongoDB pod

const uri = 'mongodb://mongo-cluster-ip-service:27017/docker-node';

Now, let's put that in terms of a ConfigMap

const uri = 'mongodb://' + \
process.env.DEV_DB_ENDPOINT + ':' + \
process.env.DEV_DB_PORT + '/' + \
process.env.DEV_DB_NAME

Create a ConfigMap

kk create cm is the easiest solution for a quick job. As always, use the -h flag to get some practical examples.

There are a few different options you can select from, but I'm just going to choose —from-literal=foo=bar, as that's the easiest for this.

Your options:

1. —from-file
2. —from-literal
3. —from-env-file

Let's create a configmap called cm1

kk create cm cm1 --from-literal=name=shannon --from-literal=config=map

This created a configmap with two keys: name and config. Thus, there are two values: shannon and map.

Just to make sure:

• Name → cm1
• key(name):value(shannon)
• key(config):value(map)

Let's confirm they were created.

kk get cm

Take a peek at the yaml too, so you can have some self-documenting code later if you want it.

kk get cm cm1 -o yaml

Let's also confirm the key value pairs are in the configmap. The output format is, in my opinion, a little funky so keep in mind a dash is added between the key and value.

kk describe cm cm1
---
Name:         cm1
Namespace:    default
Labels:       <none>
Annotations:  <none>

Data
====
config:
----
map
name:
----
shannon
Events:  <none>

Now, let's create a pod and add these env variables to them. Use kk run for a quick template. I'll be slashing a bit of the details from mine to keep this neater.

apiVersion: v1
kind: Pod
metadata:
  name: p1
spec:
  containers:
  - command:
    - printenv
    image: busybox
    name: bb1

First, we'll add an individual env variable from the cm and then we'll add all of them. Make sure to delete the pod before running this again unless you've changed the name in metadata.

apiVersion: v1
kind: Pod
metadata:
  name: p1
spec:
  containers:
  - command:
    - printenv
    image: busybox
    name: bb1
    env: # need to add everything here and below
      - name: MY_NAME  # the env variable name in your pod
        valueFrom:
          configMapKeyRef:
            name: cm1  # the cm name you created (cm1)
            key: name  # the key you created (name:shannon)

Run the pod and then check to see if the environment variables was added

kk logs p1
---
PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
HOSTNAME=p1
MY_NAME=shannon
....

Now let's add all of them instead of just one. Essentially, you just use envFrom: instead of env:

apiVersion: v1
kind: Pod
metadata:
  name: p1
spec:
  containers:
  - command:
    - printenv
    image: busybox
    name: bb1
    envFrom: # need to add everything here and below
    - configMapRef:
        name: cm1

Check the logs again to see if both your environment variables are now logged

kk logs p1
---
PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
HOSTNAME=p1
config=map
name=shannon

Cleaning up

Delete the cm

kk delete cm cm1

Let's try to build the pod again and let's check what happens

kk describe po p1
---
Warning  Failed 4s   kubelet, <node>  Error: configmap "cm1" not found

If a configMap object does not exist, the pod cannot be created. If looking at all pods, a similar error will be generated.

kk get po 
---
NAME   READY   STATUS                       RESTARTS   AGE
p1     0/1     CreateContainerConfigError   0          50s