The author selected Wikimedia Foundation Inc. to receive a donation as part of the Write for DOnations program.
Apache ZooKeeper is open-source software that enables resilient and highly reliable distributed coordination. It is commonly used in distributed systems to manage configuration information, naming services, distributed synchronization, quorum, and state. In addition, distributed systems rely on ZooKeeper to implement consensus, leader election, and group management.
In this guide, you will install and configure Apache ZooKeeper 3.4.13 on Ubuntu 18.04. To achieve resilience and high availability, ZooKeeper is intended to be replicated over a set of hosts, called an ensemble. First, you will create a standalone installation of a single-node ZooKeeper server and then add in details for setting up a multi-node cluster. The standalone installation is useful in development and testing environments, but a cluster is the most practical solution for production environments.
Before you begin this installation and configuration guide, you’ll need the following:
Because ZooKeeper keeps data in memory to achieve high throughput and low latency, production systems work best with 8GB of RAM. Lower amounts of RAM may lead to JVM swapping, which could cause ZooKeeper server latency. High ZooKeeper server latency could result in issues like client session timeouts that would have an adverse impact on system functionality.
A dedicated user should run services that handle requests over a network and consume resources. This practice creates segregation and control that will improve your environment’s security and manageability. In this step, you’ll create a non-root sudo user, named zk in this tutorial, to run the ZooKeeper service.
First, log in as the non-root sudo user that you created in the prerequisites.
ssh sammy@your_server_ip
Create the user that will run the ZooKeeper service:
- sudo useradd zk -m
Passing the -m
flag to the useradd
command will create a home directory for this user. The home directory for zk will be /home/zk
by default.
Set bash
as the default shell for the zk user:
- sudo usermod --shell /bin/bash zk
Set a password for this user:
- sudo passwd zk
Next, you will add the zk user to the sudo group so it can run commands in a privileged mode:
- usermod -aG sudo zk
In terms of security, it is recommended that you allow SSH access to as few users as possible. Logging in remotely as sammy and then using su
to switch to the desired user creates a level of separation between credentials for accessing the system and running processes. You will disable SSH access for both your zk and root user in this step.
Open your sshd_config
file:
- sudo nano /etc/ssh/sshd_config
Locate the PermitRootLogin
line and set the value to no
to disable SSH access for the root user:
PermitRootLogin no
Under the PermitRootLogin
value, add a DenyUsers
line and set the value as any user who should have SSH access disabled:
DenyUsers zk
Save and exit the file and then restart the SSH daemon to activate the changes.
- sudo systemctl restart sshd
Switch to the zk user:
- su -l zk
The -l
flag invokes a login shell after switching users. A login shell resets environment variables and provides a clean start for the user.
Enter the password at the prompt to authenticate the user.
Now that you have created, configured, and logged in as the zk user, you will create a directory to store your ZooKeeper data.
ZooKeeper persists all configuration and state data to disk so it can survive a reboot. In this step, you will create a data directory that ZooKeeper will use to read and write data. You can create the data directory on the local filesystem or on a remote storage drive. This tutorial will focus on creating the data directory on your local filesystem.
Create a directory for ZooKeeper to use:
- sudo mkdir -p /data/zookeeper
Grant your zk user ownership to the directory:
- sudo chown zk:zk /data/zookeeper
chown
changes the ownership and group of the /data/zookeeper
directory so that the user zk, who belongs to the group zk, owns the data directory.
You have successfully created and configured the data directory. When you move on to configure ZooKeeper, you will specify this path as the data directory that ZooKeeper will use to store its files.
In this step, you will manually download and extract the ZooKeeper binaries to the /opt
directory. You can use the Advanced Packaging Tool, apt
, to download ZooKeeper, but it may install an older version with different features. Installing ZooKeeper manually will give you full control to choose which version you would like to use.
Since you are downloading these files manually, start by changing to the /opt
directory:
- cd /opt
From your local machine, navigate to the Apache download page. This page will automatically provide you with the mirror closest to you for the fastest download. Click the link to the suggested mirror site, then scroll down and click zookeeper/ to view the available releases. Select the version of ZooKeeper that you would like to install. This tutorial will focus on using 3.4.13. Once you select the version, right click the binary file ending with .tar.gz
and copy the link address.
From your server, use the wget
command along with the copied link to download the ZooKeeper binaries:
- sudo wget http://apache.osuosl.org/zookeeper/zookeeper-3.4.13/zookeeper-3.4.13.tar.gz
Extract the binaries from the compressed archive:
- sudo tar -xvf zookeeper-3.4.13.tar.gz
The .tar.gz
extension represents a combination of TAR packaging followed by a GNU zip (gzip) compression. You will notice that you passed the flag -xvf
to the command to extract the archive. The flag x
stands for extract, v
enables verbose mode to show the extraction progress, and f
allows specifying the input, in our case zookeeper-3.4.13.tar.gz
, as opposed to STDIN.
Next, give the zk user ownership of the extracted binaries so that it can run the executables. You can change ownership like so:
- sudo chown zk:zk -R zookeeper-3.4.13
Next, you will configure a symbolic link to ensure that your ZooKeeper directory will remain relevant across updates. You can also use symbolic links to shorten directory names, which can lessen the time it takes to set up your configuration files.
Create a symbolic link using the ln
command.
- sudo ln -s zookeeper-3.4.13 zookeeper
Change the ownership of that link to zk:zk
. Notice that you have passed a -h
flag to change the ownership of the link itself. Not specifying -h
changes the ownership of the target of the link, which you explicitly did in the previous step.
- sudo chown -h zk:zk zookeeper
With the symbolic links created, your directory paths in the configurations will remain relevant and unchanged through future upgrades. You can now configure ZooKeeper.
Now that you’ve set up your environment, you are ready to configure ZooKeeper.
The configuration file will live in the /opt/zookeeper/conf
directory. This directory contains a sample configuration file that comes with the ZooKeeper distribution. This sample file, named zoo_sample.cfg
, contains the most common configuration parameter definitions and sample values for these parameters. Some of the common parameters are as follows:
tickTime
: Sets the length of a tick in milliseconds. A tick is a time unit used by ZooKeeper to measure the length between heartbeats. Minimum session timeouts are twice the tickTime.dataDir
: Specifies the directory used to store snapshots of the in-memory database and the transaction log for updates. You could choose to specify a separate directory for transaction logs.clientPort
: The port used to listen for client connections.maxClientCnxns
: Limits the maximum number of client connections.Create a configuration file named zoo.cfg
at /opt/zookeeper/conf
. You can create and open a file using nano
or your favorite editor:
- nano /opt/zookeeper/conf/zoo.cfg
Add the following set of properties and values to that file:
tickTime=2000
dataDir=/data/zookeeper
clientPort=2181
maxClientCnxns=60
A tickTime
of 2000 milliseconds is the suggested interval between heartbeats. A shorter interval could lead to system overhead with limited benefits. The dataDir
parameter points to the path defined by the symbolic link you created in the previous section. Conventionally, ZooKeeper uses port 2181
to listen for client connections. In most situations, 60 allowed client connections are plenty for development and testing.
Save the file and exit the editor.
You have configured ZooKeeper and are ready to start the server.
You’ve configured all the components needed to run ZooKeeper. In this step, you will start the ZooKeeper service and test your configuration by connecting to the service locally.
Navigate back to the /opt/zookeeper
directory.
- cd /opt/zookeeper
Start ZooKeeper with the zkServer.sh
command.
- bin/zkServer.sh start
You will see the following on your standard output:
OutputZooKeeper JMX enabled by default
Using config: /opt/zookeeper/bin/../conf/zoo.cfg
Starting zookeeper ... STARTED
Connect to the local ZooKeeper server with the following command:
- bin/zkCli.sh -server 127.0.0.1:2181
You will get a prompt with the label CONNECTED
. This confirms that you have a successful local, standalone ZooKeeper installation. If you encounter errors, you will want to verify that the configuration is correct.
OutputConnecting to 127.0.0.1:2181
...
...
[zk: 127.0.0.1:2181(CONNECTED) 0]
Type help
on this prompt to get a list of commands that you can execute from the client. The output will be as follows:
Output[zk: 127.0.0.1:2181(CONNECTED) 0] help
ZooKeeper -server host:port cmd args
stat path [watch]
set path data [version]
ls path [watch]
delquota [-n|-b] path
ls2 path [watch]
setAcl path acl
setquota -n|-b val path
history
redo cmdno
printwatches on|off
delete path [version]
sync path
listquota path
rmr path
get path [watch]
create [-s] [-e] path data acl
addauth scheme auth
quit
getAcl path
close
connect host:port
After you’ve done some testing, you will close the client session by typing quit
on the prompt. The ZooKeeper service will continue running after you closed the client session. Shut down the ZooKeeper service, as you’ll configure it as a systemd
service in the next step:
- bin/zkServer.sh stop
You have now installed, configured, and tested a standalone ZooKeeper service. This setup is useful to familiarize yourself with ZooKeeper, but is also helpful for developmental and testing environments. Now that you know the configuration works, you will configure systemd
to simplify the management of your ZooKeeper service.
The systemd
, system and service manager, is an init system used to bootstrap the user space and to manage system processes after boot. You can create a daemon for starting and checking the status of ZooKeeper using systemd
.
Systemd Essentials is a great introductory resource for learning more about systemd
and its constituent components.
Use your editor to create a .service
file named zk.service
at /etc/systemd/system/
.
- sudo nano /etc/systemd/system/zk.service
Add the following lines to the file to define the ZooKeeper Service:
[Unit]
Description=Zookeeper Daemon
Documentation=http://zookeeper.apache.org
Requires=network.target
After=network.target
[Service]
Type=forking
WorkingDirectory=/opt/zookeeper
User=zk
Group=zk
ExecStart=/opt/zookeeper/bin/zkServer.sh start /opt/zookeeper/conf/zoo.cfg
ExecStop=/opt/zookeeper/bin/zkServer.sh stop /opt/zookeeper/conf/zoo.cfg
ExecReload=/opt/zookeeper/bin/zkServer.sh restart /opt/zookeeper/conf/zoo.cfg
TimeoutSec=30
Restart=on-failure
[Install]
WantedBy=default.target
The Service
section in the unit file configuration specifies the working directory, the user under which the service would run, and the executable commands to start, stop, and restart the ZooKeeper service. For additional information on all the unit file configuration options, you can read the Understanding Systemd Units and Unit Files article.
Save the file and exit the editor.
Now that your systemd
configuration is in place, you can start the service:
- sudo systemctl start zk
Once you’ve confirmed that your systemd
file can successfully start the service, you will enable the service to start on boot.
- sudo systemctl enable zk
This output confirms the creation of the symbolic link:
OutputCreated symlink /etc/systemd/system/multi-user.target.wants/zk.service → /etc/systemd/system/zk.service.
Check the status of the ZooKeeper service using:
- sudo systemctl status zk
Stop the ZooKeeper service using systemctl
.
- sudo systemctl stop zk
Finally, to restart the daemon, use the following command:
- sudo systemctl restart zk
The systemd
mechanism is becoming the init system of choice on many Linux distributions. Now that you’ve configured systemd
to manage ZooKeeper, you can leverage this fast and flexible init model to start, stop, and restart the ZooKeeper service.
While the standalone ZooKeeper server is useful for development and testing, every production environment should have a replicated multi-node cluster.
Nodes in the ZooKeeper cluster that work together as an application form a quorum. Quorum refers to the minimum number of nodes that need to agree on a transaction before it’s committed. A quorum needs an odd number of nodes so that it can establish a majority. An even number of nodes may result in a tie, which would mean the nodes would not reach a majority or consensus.
In a production environment, you should run each ZooKeeper node on a separate host. This prevents service disruption due to host hardware failure or reboots. This is an important and necessary architectural consideration for building a resilient and highly available distributed system.
In this tutorial, you will install and configure three nodes in the quorum to demonstrate a multi-node setup. Before you configure a three-node cluster, you will spin up two additional servers with the same configuration as your standalone ZooKeeper installation. Ensure that the two additional nodes meet the prerequisites, and then follow steps one through six to set up a running ZooKeeper instance.
Once you’ve followed steps one through six for the new nodes, open zoo.cfg
in the editor on each node.
- sudo nano /opt/zookeeper/conf/zoo.cfg
All nodes in a quorum will need the same configuration file. In your zoo.cfg
file on each of the three nodes, add the additional configuration parameters and values for initLimit
, syncLimit
, and the servers in the quorum, at the end of the file.
tickTime=2000
dataDir=/data/zookeeper
clientPort=2181
maxClientCnxns=60
initLimit=10
syncLimit=5
server.1=your_zookeeper_node_1:2888:3888
server.2=your_zookeeper_node_2:2888:3888
server.3=your_zookeeper_node_3:2888:3888
initLimit
specifies the time that the initial synchronization phase can take. This is the time within which each of the nodes in the quorum needs to connect to the leader. syncLimit
specifies the time that can pass between sending a request and receiving an acknowledgment. This is the maximum time nodes can be out of sync from the leader. ZooKeeper nodes use a pair of ports, :2888
and :3888
, for follower nodes to connect to the leader node and for leader election, respectively.
Once you’ve updated the file on each node, you will save and exit the editor.
To complete your multi-node configuration, you will specify a node ID on each of the servers. To do this, you will create a myid
file on each node. Each file will contain a number that correlates to the server number assigned in the configuration file.
On your_zookeeper_node_1, create the myid
file that will specify the node ID:
- sudo nano /data/zookeeper/myid
Since your_zookeeper_node_1 is identified as server.1
, you will enter 1
to define the node ID. After adding the value, your file will look like this:
your_zookeeper_node_1 /data/zookeeper/myid1
Follow the same steps for the remaining nodes. The myid
file on each node should be as follows:
your_zookeeper_node_1 /data/zookeeper/myid1
your_zookeeper_node_2 /data/zookeeper/myid][environment second2
your_zookeeper_node_3 /data/zookeeper/myid][environment third3
You have now configured a three-node ZooKeeper cluster. Next, you will run the cluster and test your installation.
With each node configured to work as a cluster, you are ready to start a quorum. In this step, you will start the quorum on each node and then test your cluster by creating sample data in ZooKeeper.
To start a quorum node, first change to the /opt/zookeeper
directory on each node:
- cd /opt/zookeeper
Start each node with the following command:
- java -cp zookeeper-3.4.13.jar:lib/log4j-1.2.17.jar:lib/slf4j-log4j12-1.7.25.jar:lib/slf4j-api-1.7.25.jar:conf org.apache.zookeeper.server.quorum.QuorumPeerMain conf/zoo.cfg
As nodes start up, you will intermittently see some connection errors followed by a stage where they join the quorum and elect a leader among themselves. After a few seconds of initialization, you can start testing your installation.
Log in via SSH to your_zookeeper_node_3 as the non-root user you configured in the prerequisites:
- ssh sammy@your_zookeeper_node_3
Once logged in, switch to your zk user:
- your_zookeeper_node_3 /data/zookeeper/myid][environment thirdsu -l zk
Enter the password for the zk user. Once logged in, change the directory to /opt/zookeeper
:
- your_zookeeper_node_3 /data/zookeeper/myid][environment thirdcd /opt/zookeeper
You will now start a ZooKeeper command line client and connect to ZooKeeper on your_zookeeper_node_1:
- your_zookeeper_node_3 /data/zookeeper/myid][environment thirdbin/zkCli.sh -server your_zookeeper_node_1:2181
In the standalone installation, both the client and server were running on the same host. This allowed you to establish a client connection with the ZooKeeper server using localhost
. Since the client and server are running on different nodes in your multi-node cluster, in the previous step you needed to specify the IP address of your_zookeeper_node_1 to connect to it.
You will see the familiar prompt with the CONNECTED
label, similar to what you saw in Step 5.
Next, you will create, list, and then delete a znode. The znodes are the fundamental abstractions in ZooKeeper that are analogous to files and directories on a file system. ZooKeeper maintains its data in a hierarchical namespace, and znodes are the data registers of this namespace.
Testing that you can successfully create, list, and then delete a znode is essential to establishing that your ZooKeeper cluster is installed and configured correctly.
Create a znode named zk_znode_1
and associate the string sample_data
with it.
- create /zk_znode_1 sample_data
You will see the following output once created:
OutputCreated /zk_znode_1
List the newly created znode:
- ls /
Get the data associated with it:
- get /zk_znode_1
ZooKeeper will respond like so:
Output[zk: your_zookeeper_node_1:2181(CONNECTED)] ls /
[zk_znode_1, zookeeper]
[zk: your_zookeeper_node_1:2181(CONNECTED)] get /zk_znode_1
sample_data
cZxid = 0x100000002
ctime = Tue Nov 06 19:47:41 UTC 2018
mZxid = 0x100000002
mtime = Tue Nov 06 19:47:41 UTC 2018
pZxid = 0x100000002
cversion = 0
dataVersion = 0
aclVersion = 0
ephemeralOwner = 0x0
dataLength = 11
numChildren = 0
The output confirms the value, sample_data
, that you associated with zk_node_1
. ZooKeeper also provides additional information about creation time, ctime
, and modification time, mtime
. ZooKeeper is a versioned data store, so it also presents you with metadata about the data version.
Delete the zk_znode_1
znode:
- delete /zk_znode_1
In this step, you successfully tested connectivity between two of your ZooKeeper nodes. You also learned basic znode management by creating, listing, and deleting znodes. Your multi-node configuration is complete, and you are ready to start using ZooKeeper.
In this tutorial, you configured and tested both a standalone and multi-node ZooKeeper environment. Now that your multi-node ZooKeeper deployment is ready to use, you can review the official ZooKeeper documentation for additional information and projects.
Thanks for learning with the DigitalOcean Community. Check out our offerings for compute, storage, networking, and managed databases.
This textbox defaults to using Markdown to format your answer.
You can type !ref in this text area to quickly search our full set of tutorials, documentation & marketplace offerings and insert the link!
Why should we do that instead of just starting (or restarting) server through bin/zkServer.sh as described above?
You must download the bin.tar.gz in order to properly bin/zkServer.sh start
I was not able to get
sudo systemctl start zk
orsudo systemctl restart
to work. However,sudo systemctl stop zk
andsudo systemctl status zk
work fine.Also,
sudo /opt/zookeeper/bin/zkServer.sh start /opt/zookeeper/conf/zoo.cfg
andsudo /opt/zookeeper/bin/zkServer.sh restart /opt/zookeeper/conf/zoo.cfg
work just fine.Any idea why that might be?
Is it necessary to create a new user in order to install zookeeper? Can I do this with an existing user?