pravega-operator

README

Pravega Operator

Project status: alpha

The project is currently alpha. While no breaking API changes are currently planned, we reserve the right to address bugs and change the API before the project is declared stable.

Table of Contents

• Overview
• Requirements
• Quickstart
• Install the Operator
• Deploying in Test Mode
• Upgrade the Operator
• Install a sample Pravega Cluster
• Scale a Pravega Cluster
• Upgrade a Pravega Cluster
• Uninstall the Pravega Cluster
• Uninstall the Operator
• Manual installation
• Configuration
• Development
• Releases
• Troubleshooting

Overview

Pravega is an open source distributed storage service implementing Streams. It offers Stream as the main primitive for the foundation of reliable storage systems: a high-performance, durable, elastic, and unlimited append-only byte stream with strict ordering and consistency.

The Pravega Operator manages Pravega clusters deployed to Kubernetes and automates tasks related to operating a Pravega cluster.

• Create and destroy a Pravega cluster
• Resize cluster
• Rolling upgrades

Requirements

• Kubernetes 1.15+
• Helm 3+
• An existing Apache Zookeeper 3.6.1 cluster. This can be easily deployed using our Zookeeper operator
• An existing Apache Bookkeeper 4.9.2 cluster. This can be easily deployed using our BookKeeper Operator

Quickstart

Install the Operator

Note: If you are running on Google Kubernetes Engine (GKE), please check this first.

We recommend using helm to deploy a Pravega Operator. Check out the helm installation document for instructions.

Deploying in Test Mode

The Operator can be run in a "test" mode if we want to create pravega on minikube or on a cluster with very limited resources by enabling testmode: true in values.yaml file. Operator running in test mode skips minimum replica requirement checks on Pravega components. "Test" mode ensures a bare minimum setup of pravega and is not recommended to be used in production environments.

Upgrade the Operator

For upgrading the pravega operator check the document operator-upgrade

Install a sample Pravega cluster

Set up Tier 2 Storage

Pravega requires a long term storage provider known as longtermStorage.

Check out the available options for long term storage and how to configure it.

For demo purposes, you can quickly install a toy NFS server.

$ helm install stable/nfs-server-provisioner --generate-name

And create a PVC for longtermStorage that utilizes it.

$ kubectl create -f ./example/pvc-tier2.yaml

Install a Pravega cluster

Use Helm to install a sample Pravega cluster with release name bar.

$ helm install bar charts/pravega --set zookeeperUri=[ZOOKEEPER_HOST] --set bookkeeperUri=[BOOKKEEPER_SVC] --set storage.longtermStorage.filesystem.pvc=[TIER2_NAME]

where:

• [ZOOKEEPER_HOST] is the host or IP address of your Zookeeper deployment (e.g. zookeeper-client:2181). Multiple Zookeeper URIs can be specified, use a comma-separated list and DO NOT leave any spaces in between (e.g. zookeeper-0:2181,zookeeper-1:2181,zookeeper-2:2181).
• [BOOKKEEPER_SVC] is the is the name of the headless service of your Bookkeeper deployment (e.g. bookkeeper-bookie-0.bookkeeper-bookie-headless.default.svc.cluster.local:3181,bookkeeper-bookie-1.bookkeeper-bookie-headless.default.svc.cluster.local:3181,bookkeeper-bookie-2.bookkeeper-bookie-headless.default.svc.cluster.local:3181).
• [TIER2_NAME] is the longtermStorage PersistentVolumeClaim name. pravega-tier2 if you created the PVC above.

Check out the Pravega Helm Chart for the complete list of configurable parameters.

Verify that the cluster instances and its components are being created.

$ kubectl get PravegaCluster
NAME          VERSION   DESIRED MEMBERS   READY MEMBERS   AGE
bar-pravega   0.4.0     7                 0               25s

After a couple of minutes, all cluster members should become ready.

$ kubectl get PravegaCluster
NAME         VERSION   DESIRED MEMBERS   READY MEMBERS   AGE
bar-pravega  0.4.0     7                 7               2m

$ kubectl get all -l pravega_cluster=bar-pravega
NAME                                          READY   STATUS    RESTARTS   AGE
pod/bar-pravega-controller-64ff87fc49-kqp9k   1/1     Running   0          2m
pod/bar-pravega-segmentstore-0                1/1     Running   0          2m
pod/bar-pravega-segmentstore-1                1/1     Running   0          1m
pod/bar-pravega-segmentstore-2                1/1     Running   0          30s

NAME                                        TYPE        CLUSTER-IP    EXTERNAL-IP   PORT(S)              AGE
service/bar-pravega-controller              ClusterIP   10.23.244.3   <none>        10080/TCP,9090/TCP   2m
service/bar-pravega-segmentstore-headless   ClusterIP   None          <none>        12345/TCP            2m

NAME                                                    DESIRED   CURRENT   READY   AGE
replicaset.apps/bar-pravega-controller-64ff87fc49       1         1         1       2m

NAME                                            DESIRED   CURRENT   AGE
statefulset.apps/bar-pravega-segmentstore       3         3         2m

By default, a PravegaCluster instance is only accessible within the cluster through the Controller ClusterIP service. From within the Kubernetes cluster, a client can connect to Pravega at:

tcp://<pravega-name>-pravega-controller.<namespace>:9090

And the REST management interface is available at:

http://<pravega-name>-pravega-controller.<namespace>:10080/

Check out the external access documentation if your clients need to connect to Pravega from outside Kubernetes.

Scale a Pravega cluster

You can scale Pravega components independently by modifying their corresponding field in the Pravega resource spec. You can either kubectl edit the cluster or kubectl patch it. If you edit it, update the number of replicas for BookKeeper, Controller, and/or Segment Store and save the updated spec.

Example of patching the Pravega resource to scale the Segment Store instances to 4.

kubectl patch PravegaCluster <pravega-name> --type='json' -p='[{"op": "replace", "path": "/spec/pravega/segmentStoreReplicas", "value": 4}]'

Upgrade a Pravega cluster

Check out the upgrade guide.

Uninstall the Pravega cluster

$ helm uninstall bar
$ kubectl delete -f ./example/pvc-tier2.yaml

Uninstall the Operator

Note that the Pravega clusters managed by the Pravega operator will NOT be deleted even if the operator is uninstalled.

$ helm uninstall foo

If you want to delete the Pravega clusters, make sure to do it before uninstalling the operator.

Manual installation

You can also manually install/uninstall the operator and Pravega with kubectl commands. Check out the manual installation document for instructions.

Configuration

Check out the configuration document.

Development

Check out the development guide.

Releases

The latest Pravega releases can be found on the Github Release project page.

Troubleshooting

Check out the troubleshooting document.