Advanced Flame k8s Adapter basead on Operator Pattern.
If available in Hex, the package can be installed
by adding flame_k8s to your list of dependencies in mix.exs:
def deps do
[
{:flame_k8s, "~> 0.1.0"}
]
endNOTE: You need to install our Kubernetes Controller in the Kubernetes where you want to run your application. Follow the instructions below
To install flame-k8s kubernetes controller, just download the manifest from the release page and apply it to your cluster.
Configure the flame backend in our configuration.
# config.exs
if config_env() == :prod do
config :flame, :backend, FLAME.K8sBackend
config :flame, FLAME.K8sBackend, log: :debug
endYou need to enable Flame in Kubernetes as well. See the example below:
# my-application.yaml
---
apiVersion: apps/v1
kind: Deployment
metadata:
name: flame-parent-example
namespace: default
spec:
replicas: 1
selector:
matchLabels:
app: flame-parent-example
template:
metadata:
annotations:
flame.org/enabled: "true"
flame.org/dist-auto-config: "true"
flame.org/otp-app: "my_app_release_name"
spec:
containers:
- image: eigr/flame-parent-example:1.1.1
name: flame-parent-example
resources:
limits:
cpu: 200m
memory: 200Mi
requests:
cpu: 200m
memory: 200MiThe most important part is:
template:
metadata:
annotations:
flame.org/enabled: "true"
flame.org/dist-auto-config: "true"
flame.org/otp-app: "my_app_release_name"See what each annotation means in the following table:
| Annotation | Default | Detail |
|---|---|---|
| flame.org/enabled | "false" | Enable or disable Flame. |
| flame.org/dist-auto-config | "false" | Auto configure RELEASE_DISTRIBUTION and RELEASE_NODE based on otp application name. |
| flame.org/otp-app | Application release name. Required if dist-auto-config is set to "true". | |
| flame.org/pool-config-ref | "default-pool" | Flame Pool configuration reference name. See more in the Configuration section. |
| flame.org/runner-termination-timeout | 60000 | Timeout in milliseconds that the Runner will have to finish before the controller sends the POD delete command. |
Now you can start scaling your applications with Flame... with a little help from eigr \0/
TODO
The Flame k8s Controller gives you the possibility to configure different runner profiles. These profiles will be used when creating PODs to run Runners in Kubernetes. To configure a new Runner Pool, simply define the following yaml file and apply it to the Kubernetes cluster.
# my-runner.yaml
---
apiVersion: flame.org/v1
kind: FlamePool
metadata:
name: my-runner-pool
namespace: default
spec:
podTemplate:
spec: # This is a pod template specification. See https://kubernetes.io/docs/concepts/workloads/pods/#pod-templates
containers:
- env:
- name: MY_VAR
value: "my-value"
resources:
limits:
cpu: 200m
memory: 1Gi
requests:
cpu: 200m
memory: 2Gi
volumeMounts:
- mountPath: /app/.cache/bakeware/
name: bakeware-cache
volumes:
- name: bakeware-cache
emptyDir: {}Then:
kubectl apply -f my-runner.yamlOnce this is done, simply add the annotation flame.org/pool-config-ref to your Deployment file. Example:
# my-application.yaml
---
apiVersion: apps/v1
kind: Deployment
metadata:
name: flame-parent-example
namespace: default
spec:
replicas: 1
selector:
matchLabels:
app: flame-parent-example
template:
metadata:
annotations:
flame.org/enabled: "true"
flame.org/pool-config-ref: "my-runner-pool"
spec:
containers:
- image: eigr/flame-parent-example:1.1.1
... You can also list all Runner pools configured on the system with the command:
kubectl --all-namespaces get poolsThe controller also defines a default Pool which in turn has the following configuration:
---
apiVersion: flame.org/v1
kind: FlamePool
metadata:
name: default-pool
namespace: flame
spec:
podTemplate:
spec:
containers:
- env:
- name: PHX_SERVER
value: "false"
- name: MIX_ENV
value: prod
- name: POD_NAME
valueFrom:
fieldRef:
fieldPath: metadata.name
- name: POD_NAMESPACE
valueFrom:
fieldRef:
fieldPath: metadata.namespace
- name: POD_IP
valueFrom:
fieldRef:
fieldPath: status.podIP
# Other vars...
resources:
requests:
cpu: 50m
memory: 128Mi
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent