@@ -31,28 +31,29 @@ You should already have turned up a Kubernetes cluster. To get the most of this
...
@@ -31,28 +31,29 @@ You should already have turned up a Kubernetes cluster. To get the most of this
The Celery task queue will need to communicate with the RabbitMQ broker. RabbitMQ will eventually appear on a separate pod, but since pods are ephemeral we need a service that can transparently route requests to RabbitMQ.
The Celery task queue will need to communicate with the RabbitMQ broker. RabbitMQ will eventually appear on a separate pod, but since pods are ephemeral we need a service that can transparently route requests to RabbitMQ.
Use the file `examples/celery-rabbitmq/rabbitmq-service.json`:
Use the file `examples/celery-rabbitmq/rabbitmq-service.yaml`:
**NOTE**: If you're running Kubernetes from source, you can use `cluster/kubectl.sh` instead of `kubectl`.
**NOTE**: If you're running Kubernetes from source, you can use `cluster/kubectl.sh` instead of `kubectl`.
...
@@ -62,81 +63,71 @@ This service allows other pods to connect to the rabbitmq. To them, it will be s
...
@@ -62,81 +63,71 @@ This service allows other pods to connect to the rabbitmq. To them, it will be s
## Step 2: Fire up RabbitMQ
## Step 2: Fire up RabbitMQ
A RabbitMQ broker can be turned up using the file `examples/celery-rabbitmq/rabbitmq-controller.json`:
A RabbitMQ broker can be turned up using the file `examples/celery-rabbitmq/rabbitmq-controller.yaml`:
```js
```yaml
{
apiVersion:v1beta3
"id":"rabbitmq-controller",
kind:ReplicationController
"kind":"ReplicationController",
metadata:
"apiVersion":"v1beta1",
labels:
"desiredState":{
name:rabbitmq
"replicas":1,
name:rabbitmq-controller
"replicaSelector":{"name":"rabbitmq"},
spec:
"podTemplate":{
replicas:1
"desiredState":{
selector:
"manifest":{
component:rabbitmq
"version":"v1beta1",
template:
"id":"rabbitmq",
metadata:
"containers":[{
labels:
"name":"rabbitmq",
app:taskQueue
"image":"library/rabbitmq",
component:rabbitmq
"cpu":100,
spec:
"ports":[{"containerPort":5672,"hostPort":5672}]
containers:
}]
-image:rabbitmq
}
name:rabbitmq
},
ports:
"labels":{
-containerPort:5672
"name":"rabbitmq",
protocol:TCP
"app":"taskQueue"
resources:
}
limits:
}
cpu:100m
},
"labels":{
"name":"rabbitmq"
}
}
```
```
Running `$ kubectl create -f examples/celery-rabbitmq/rabbitmq-controller.json` brings up a replication controller that ensures one pod exists which is running a RabbitMQ instance.
Running `$ kubectl create -f examples/celery-rabbitmq/rabbitmq-controller.yaml` brings up a replication controller that ensures one pod exists which is running a RabbitMQ instance.
Note that bringing up the pod includes pulling down a docker image, which may take a few moments. This applies to all other pods in this example.
Note that bringing up the pod includes pulling down a docker image, which may take a few moments. This applies to all other pods in this example.
## Step 3: Fire up Celery
## Step 3: Fire up Celery
Bringing up the celery worker is done by running `$ kubectl create -f examples/celery-rabbitmq/celery-controller.json`, which contains this:
Bringing up the celery worker is done by running `$ kubectl create -f examples/celery-rabbitmq/celery-controller.yaml`, which contains this:
```js
```yaml
{
apiVersion:v1beta3
"id":"celery-controller",
kind:ReplicationController
"kind":"ReplicationController",
metadata:
"apiVersion":"v1beta1",
labels:
"desiredState":{
name:celery
"replicas":1,
name:celery-controller
"replicaSelector":{"name":"celery"},
spec:
"podTemplate":{
replicas:1
"desiredState":{
selector:
"manifest":{
component:celery
"version":"v1beta1",
template:
"id":"celery",
metadata:
"containers":[{
labels:
"name":"celery",
app:taskQueue
"image":"endocode/celery-app-add",
component:celery
"cpu":100,
spec:
"ports":[{"containerPort":5672,"hostPort":5672}]
containers:
}]
-image:endocode/celery-app-add
}
name:celery
},
ports:
"labels":{
-containerPort:5672
"name":"celery",
protocol:TCP
"app":"taskQueue"
resources:
}
limits:
}
cpu:100m
},
"labels":{
"name":"celery"
}
}
```
```
There are several things to point out here...
There are several things to point out here...
...
@@ -187,39 +178,35 @@ The question now is, how do you see what's going on?
...
@@ -187,39 +178,35 @@ The question now is, how do you see what's going on?
Flower is a web-based tool for monitoring and administrating Celery clusters. By connecting to the node that contains Celery, you can see the behaviour of all the workers and their tasks in real-time.
Flower is a web-based tool for monitoring and administrating Celery clusters. By connecting to the node that contains Celery, you can see the behaviour of all the workers and their tasks in real-time.
To bring up the frontend, run this command `$ kubectl create -f examples/celery-rabbitmq/celery-controller.json`. This controller is defined as so:
To bring up the frontend, run this command `$ kubectl create -f examples/celery-rabbitmq/flower-controller.yaml`. This controller is defined as so:
```js
```yaml
{
apiVersion:v1beta3
"id":"flower-controller",
kind:ReplicationController
"kind":"ReplicationController",
metadata:
"apiVersion":"v1beta1",
labels:
"desiredState":{
name:flower
"replicas":1,
name:flower-controller
"replicaSelector":{"name":"flower"},
spec:
"podTemplate":{
replicas:1
"desiredState":{
selector:
"manifest":{
component:flower
"version":"v1beta1",
template:
"id":"flower",
metadata:
"containers":[{
labels:
"name":"flower",
app:taskQueue
"image":"endocode/flower",
component:flower
"cpu":100,
spec:
"ports":[{"containerPort":5555,"hostPort":5555}]
containers:
}]
-image:endocode/flower
}
name:flower
},
ports:
"labels":{
-containerPort:5555
"name":"flower",
hostPort:5555
"app":"taskQueue"
protocol:TCP
}
resources:
}
limits:
},
cpu:100m
"labels":{
"name":"flower"
}
}
```
```
This will bring up a new pod with Flower installed and port 5555 (Flower's default port) exposed. This image uses the following command to start Flower:
This will bring up a new pod with Flower installed and port 5555 (Flower's default port) exposed. This image uses the following command to start Flower:
