14 KiB
Cloud Native Deployments of Cassandra using Kubernetes
The following document describes the development of a cloud native Cassandra deployment on Kubernetes. When we say cloud native we mean an application which understands that it is running within a cluster manager, and uses this cluster management infrastructure to help implement the application. In particular, in this instance, a custom Cassandra SeedProvider is used to enable Cassandra to dynamically discover new Cassandra nodes as they join the cluster.
This document also attempts to describe the core components of Kubernetes, Pods, Services and Replication Controllers.
Prerequisites
This example assumes that you have a Kubernetes cluster installed and running, and that you have installed the kubectl command line tool somewhere in your path. Please see the getting started for installation instructions for your platform.
A note for the impatient
This is a somewhat long tutorial. If you want to jump straight to the "do it now" commands, please see the tl; dr at the end.
Simple Single Pod Cassandra Node
In Kubernetes, the atomic unit of an application is a Pod. A Pod is one or more containers that must be scheduled onto the same host. All containers in a pod share a network namespace, and may optionally share mounted volumes. In this simple case, we define a single container running Cassandra for our pod:
apiVersion: v1beta3
kind: Pod
metadata:
labels:
name: cassandra
name: cassandra
spec:
containers:
- args:
- /run.sh
resources:
limits:
cpu: "1"
image: kubernetes/cassandra:v2
name: cassandra
ports:
- name: cql
containerPort: 9042
- name: thrift
containerPort: 9160
volumeMounts:
- name: data
mountPath: /cassandra_data
env:
- name: MAX_HEAP_SIZE
value: 512M
- name: HEAP_NEWSIZE
value: 100M
- name: KUBERNETES_API_PROTOCOL
value: http
volumes:
- name: data
emptyDir: {}
There are a few things to note in this description. First is that we are running the kubernetes/cassandra image. This is a standard Cassandra installation on top of Debian. However it also adds a custom SeedProvider to Cassandra. In Cassandra, a SeedProvider bootstraps the gossip protocol that Cassandra uses to find other nodes. The KubernetesSeedProvider discovers the Kubernetes API Server using the built in Kubernetes discovery service, and then uses the Kubernetes API to find new nodes (more on this later)
You may also note that we are setting some Cassandra parameters (MAX_HEAP_SIZE and HEAP_NEWSIZE). We also tell Kubernetes that the container exposes both the CQL and Thrift API ports. Finally, we tell the cluster manager that we need 1 cpu (1 core).
Given this configuration, we can create the pod as follows
$ kubectl create -f cassandra.yaml
After a few moments, you should be able to see the pod running:
$ kubectl get pods cassandra
POD IP CONTAINER(S) IMAGE(S) HOST LABELS STATUS CREATED MESSAGE
cassandra 10.244.3.3 kubernetes-minion-sft2/104.197.42.181 name=cassandra Running 21 seconds
cassandra kubernetes/cassandra:v2 Running 3 seconds
Adding a Cassandra Service
In Kubernetes a Service describes a set of Pods that perform the same task. For example, the set of nodes in a Cassandra cluster, or even the single node we created above. An important use for a Service is to create a load balancer which distributes traffic across members of the set. But a Service can also be used as a standing query which makes a dynamically changing set of Pods (or the single Pod we've already created) available via the Kubernetes API. This is the way that we use initially use Services with Cassandra.
Here is the service description:
apiVersion: v1beta3
kind: Service
metadata:
labels:
name: cassandra
name: cassandra
spec:
ports:
- port: 9042
targetPort: 9042
selector:
name: cassandra
The important thing to note here is the selector. It is a query over labels, that identifies the set of Pods contained by the Service. In this case the selector is name=cassandra. If you look back at the Pod specification above, you'll see that the pod has the corresponding label, so it will be selected for membership in this Service.
Create this service as follows:
$ kubectl create -f cassandra-service.yaml
Once the service is created, you can query it's endpoints:
$ kubectl get endpoints cassandra -o yaml
apiVersion: v1beta3
kind: Endpoints
metadata:
creationTimestamp: 2015-04-23T17:21:27Z
name: cassandra
namespace: default
resourceVersion: "857"
selfLink: /api/v1beta3/namespaces/default/endpoints/cassandra
uid: 2c7d36bf-e9dd-11e4-a7ed-42010af011dd
subsets:
- addresses:
- IP: 10.244.3.3
targetRef:
kind: Pod
name: cassandra
namespace: default
resourceVersion: "769"
uid: d185872c-e9dc-11e4-a7ed-42010af011dd
ports:
- port: 9042
protocol: TCP
You can see that the Service has found the pod we created in step one.
Adding replicated nodes
Of course, a single node cluster isn't particularly interesting. The real power of Kubernetes and Cassandra lies in easily building a replicated, resizable Cassandra cluster.
In Kubernetes a Replication Controller is responsible for replicating sets of identical pods. Like a Service it has a selector query which identifies the members of it's set. Unlike a Service it also has a desired number of replicas, and it will create or delete Pods to ensure that the number of Pods matches up with it's desired state.
Replication Controllers will "adopt" existing pods that match their selector query, so let's create a Replication Controller with a single replica to adopt our existing Cassandra Pod.
apiVersion: v1beta3
kind: ReplicationController
metadata:
labels:
name: cassandra
name: cassandra
spec:
replicas: 1
selector:
name: cassandra
template:
metadata:
labels:
name: cassandra
spec:
containers:
- command:
- /run.sh
resources:
limits:
cpu: 1
env:
- name: MAX_HEAP_SIZE
key: MAX_HEAP_SIZE
value: 512M
- name: HEAP_NEWSIZE
key: HEAP_NEWSIZE
value: 100M
image: "kubernetes/cassandra:v2"
name: cassandra
ports:
- containerPort: 9042
name: cql
- containerPort: 9160
name: thrift
volumeMounts:
- mountPath: /cassandra_data
name: data
volumes:
- name: data
emptyDir: {}
The bulk of the replication controller config is actually identical to the Cassandra pod declaration above, it simply gives the controller a recipe to use when creating new pods. The other parts are the replicaSelector which contains the controller's selector query, and the replicas parameter which specifies the desired number of replicas, in this case 1.
Create this controller:
$ kubectl create -f cassandra-controller.yaml
Now this is actually not that interesting, since we haven't actually done anything new. Now it will get interesting.
Let's resize our cluster to 2:
$ kubectl resize rc cassandra --replicas=2
Now if you list the pods in your cluster, you should see two cassandra pods:
$ kubectl get pods
POD IP CONTAINER(S) IMAGE(S) HOST LABELS STATUS CREATED MESSAGE
cassandra 10.244.3.3 kubernetes-minion-sft2/104.197.42.181 name=cassandra Running 7 minutes
cassandra kubernetes/cassandra:v2 Running 7 minutes
cassandra-gnhk8 10.244.0.5 kubernetes-minion-dqz3/104.197.2.71 name=cassandra Running About a minute
cassandra kubernetes/cassandra:v2 Running 51 seconds
Notice that one of the pods has the human readable name cassandra that you specified in your config before, and one has a random string, since it was named by the replication controller.
To prove that this all works, you can use the nodetool command to examine the status of the cluster, for example:
$ ssh 104.197.42.181
$ docker exec <cassandra-container-id> nodetool status
Datacenter: datacenter1
=======================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
-- Address Load Tokens Owns (effective) Host ID Rack
UN 10.244.0.5 74.09 KB 256 100.0% 86feda0f-f070-4a5b-bda1-2eeb0ad08b77 rack1
UN 10.244.3.3 51.28 KB 256 100.0% dafe3154-1d67-42e1-ac1d-78e7e80dce2b rack1
Now let's resize our cluster to 4 nodes:
$ kubectl resize rc cassandra --replicas=4
Examining the status again:
$ docker exec <cassandra-container-id> nodetool status
Datacenter: datacenter1
=======================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
-- Address Load Tokens Owns (effective) Host ID Rack
UN 10.244.2.3 57.61 KB 256 49.1% 9d560d8e-dafb-4a88-8e2f-f554379c21c3 rack1
UN 10.244.1.7 41.1 KB 256 50.2% 68b8cc9c-2b76-44a4-b033-31402a77b839 rack1
UN 10.244.0.5 74.09 KB 256 49.7% 86feda0f-f070-4a5b-bda1-2eeb0ad08b77 rack1
UN 10.244.3.3 51.28 KB 256 51.0% dafe3154-1d67-42e1-ac1d-78e7e80dce2b rack1
tl; dr;
For those of you who are impatient, here is the summary of the commands we ran in this tutorial.
# create a single cassandra node
kubectl create -f cassandra.yaml
# create a service to track all cassandra nodes
kubectl create -f cassandra-service.yaml
# create a replication controller to replicate cassandra nodes
kubectl create -f cassandra-controller.yaml
# scale up to 2 nodes
kubectl resize rc cassandra --replicas=2
# validate the cluster
docker exec <container-id> nodetool status
# scale up to 4 nodes
kubectl resize rc cassandra --replicas=4
Seed Provider Source
package io.k8s.cassandra;
import java.io.IOException;
import java.net.InetAddress;
import java.net.UnknownHostException;
import java.net.URL;
import java.net.URLConnection;
import java.security.KeyManagementException;
import java.security.NoSuchAlgorithmException;
import java.security.SecureRandom;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import org.codehaus.jackson.JsonNode;
import org.codehaus.jackson.annotate.JsonIgnoreProperties;
import org.codehaus.jackson.map.ObjectMapper;
import org.apache.cassandra.locator.SeedProvider;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
public class KubernetesSeedProvider implements SeedProvider {
@JsonIgnoreProperties(ignoreUnknown = true)
static class Endpoints {
public String[] endpoints;
}
private static String getEnvOrDefault(String var, String def) {
String val = System.getenv(var);
if (val == null) {
val = def;
}
return val;
}
private static final Logger logger = LoggerFactory.getLogger(KubernetesSeedProvider.class);
private List defaultSeeds;
public KubernetesSeedProvider(Map<String, String> params) {
// Taken from SimpleSeedProvider.java
// These are used as a fallback, if we get nothing from k8s.
String[] hosts = params.get("seeds").split(",", -1);
defaultSeeds = new ArrayList<InetAddress>(hosts.length);
for (String host : hosts)
{
try {
defaultSeeds.add(InetAddress.getByName(host.trim()));
}
catch (UnknownHostException ex)
{
// not fatal... DD will bark if there end up being zero seeds.
logger.warn("Seed provider couldn't lookup host " + host);
}
}
}
public List<InetAddress> getSeeds() {
List<InetAddress> list = new ArrayList<InetAddress>();
String protocol = getEnvOrDefault("KUBERNETES_API_PROTOCOL", "http");
String hostName = getEnvOrDefault("KUBERNETES_RO_SERVICE_HOST", "localhost");
String hostPort = getEnvOrDefault("KUBERNETES_RO_SERVICE_PORT", "8080");
String host = protocol + "://" + hostName + ":" + hostPort;
String serviceName = getEnvOrDefault("CASSANDRA_SERVICE", "cassandra");
String path = "/api/v1beta3/endpoints/";
try {
URL url = new URL(host + path + serviceName);
ObjectMapper mapper = new ObjectMapper();
Endpoints endpoints = mapper.readValue(url, Endpoints.class);
if (endpoints != null) {
// Here is a problem point, endpoints.endpoints can be null in first node cases.
if (endpoints.endpoints != null){
for (String endpoint : endpoints.endpoints) {
String[] parts = endpoint.split(":");
list.add(InetAddress.getByName(parts[0]));
}
}
}
} catch (IOException ex) {
logger.warn("Request to kubernetes apiserver failed");
}
if (list.size() == 0) {
// If we got nothing, we might be the first instance, in that case
// fall back on the seeds that were passed in cassandra.yaml.
return defaultSeeds;
}
return list;
}
// Simple main to test the implementation
public static void main(String[] args) {
SeedProvider provider = new KubernetesSeedProvider(new HashMap<String, String>());
System.out.println(provider.getSeeds());
}
}