Skip to content

Manual Installation

This page describes the prerequisites and installation steps to deploy Pravega in a multi-node production environment.


Cloud Storage

Pravega supports Amazon S3, Azure Blob Storage, Google Cloud Storage, HDFS and mounted distributed filesystems for Long Term Storage. With the S3 protocol being widely supported by 3rd party storage systems, Pravega can be run either alongside a private storage cluster or by targeting public cloud resources.


If it is easier to mount an NFS share, then FILESYSTEM can be used for Long Term Storage. The following configuration options are necessary to configure the FILESYSTEM as Long Term Storage. = FILESYSTEM
filesystem.root = /mnt/tier2
where /mnt/tier2 is replaced with your nfs share and FILESYSTEM is a keyword.

Other options for include AZURE, EXTENDEDS3, GCP, HDFS, and S3.


Install Java 11 or later. Packages are available for all major operating systems.


Pravega requires Zookeeper 3.6.3. At least 3 Zookeeper nodes are recommended for a quorum. No special configuration is required for Zookeeper but it is recommended to use a dedicated cluster for Pravega.

This specific version of Zookeeper can be downloaded from Apache at apache-zookeeper-3.6.3.tar.gz.

For installing Zookeeper see the Getting Started Guide.


Pravega requires Bookkeeper 4.14.1. At least 3 Bookkeeper servers are recommended for a quorum.

This specific version of Bookkeeper can be downloaded from Apache at bookkeeper-server-4.14.1-bin.tar.gz.

For installing Bookkeeper see the Getting Started Guide. Some specific Pravega instructions are shown below. All sets are assumed to be run from the bookkeeper-server-4.14.1 directory.

Bookkeeper Configuration

In the file conf/bk_server.conf, the following configuration options should be implemented:


# Alternatively specify a different path to the storage for /bk

Initializing Zookeeper paths

The following paths need to be created in Zookeeper. Open the zookeeper-3.6.3 directory on the Zookeeper servers and run the following paths:

bin/ -server $ZK_URL create /pravega
bin/ -server $ZK_URL create /pravega/bookkeeper
Replace <$ZK_URL> with the IP address of the Zookeeper nodes.

Running Bookkeeper

The bookie needs the following formatting before starting it:

bin/bookkeeper shell metaformat -nonInteractive

Start the bookie as mentioned below:

bin/bookkeeper bookie

Running Bookkeeper with encryption enabled

Apache BookKeeper can be deployed with TLS enabled. Details can be found here.

Installing Pravega

For non-production systems, the containers can be used that are provided by the Docker installation to run non-production HDFS, Zookeeper or Bookkeeper.

The following two key components of Pravega needs to be run:

  • Controller: The Control plane for Pravega. Installation requires at least one Controller. (Two or more are recommended for HA).

  • Segment Store: The Storage node for Pravega. Installation requires at least one Segment Store.

Before we start, the latest Pravega release needs to be downloaded from the GitHub Releases page.


For a simple 3 node cluster, the following table depicts on layout of the services:

Node 1 Node 2 Node 3
Zookeeper X X X
Bookkeeper X X X
Pravega Controller X X
Pravega Segment Store X X X

All Nodes

On each node, extract the distribution package to the desired directory as follows:

tar xfvz pravega-<version>.tgz
cd pravega-<version>

Installation of the Controller

The controller can be run by using the following command. Replace <zk-ip> with the IP address of the Zookeeper nodes in the following command:

ZK_URL=<zk-ip>:2181 bin/pravega-controller

Instead specifying the <zk-ip> on every startup, we can edit the conf/controller.conf file and change the zk url as follows:

    zk {
      url = "<zk-ip>:2181"

Then run the controller with the following command:


Installation of the Segment Store

In the file conf/, make the following changes as mentioned: Replace <zk-ip> and <controller-ip> with the IPs of the respective services.


# Settings required for your Long Term Storage provider

After making the configuration changes, the segment store can be run using the following command:


Running a Pravega Cluster with Security enabled

Steps for securing a distributed mode cluster can be found here.

For detailed information about security configuration parameters for Controller) and Segment Store, see [this]((../security/ document.