Deploy Redis on your VPS


Install Redis

Prepare System for Redis

Issue the following commands to update your system's package repositories and ensure that all installed packages are up to date:

apt-get update apt-get upgrade 

Install required prerequisites with the following command:

apt-get install build-essential 

This guide only provides instructions for installing and managing Redis itself. The application you deploy likely requires additional infrastructure, dependencies, and utilities.

Download and Compile Software

Begin the installation process by issuing the following sequence of commands to download the software and prepare it for use:

cd /opt/ wget tar -zxvf /opt/redis-2.0.4.tar.gz mv /opt/redis-2.0.4/ /opt/redis/ cd /opt/redis make 

This will download and compile the 2.0.4 version of Redis. Check the Redis upstream project sourceExternal Link to ensure that you are downloading the most up to date version of Redis. It is important to use the most up to date version of the software to avoid security flaws and bugs as well as to take advantage of the latest features. When you compile software manually, you are responsible for ensuring your system is running the most current version without the assistance of your system's package management tools.

Redis Configuration

All Redis configuration options can be specified in the redis.conf file located at /opt/redis/redis.conf. Issue the following command to create a backup of the default configuration file:

mv /opt/redis/redis.conf /opt/redis/redis.conf.default 

Consider the following configuration:

File: redis.conf

daemonize yes pidfile /var/run/ logfile /var/log/redis.log  port 6379 bind timeout 300  loglevel notice  ## Default configuration options databases 16  save 900 1 save 300 10 save 60 10000  rdbcompression yes dbfilename dump.rdb  dir /opt/redis/ appendonly no  glueoutputbuf yes 

Most of the values in this configuration mirror the default Redis configuration. However, this configuration configures Redis to run in a daemon mode bound only to the local network interface. You may want to change these values depending on the needs of your application.

Monitor for Software Updates and Security Notices

When running software compiled or installed directly from sources provided by upstream developers, you are responsible for monitoring updates, bug fixes, and security issues. After becoming aware of releases and potential issues, update your software to resolve flaws and prevent possible system compromise. Monitoring releases and maintaining up to date versions of all software is crucial for the security and integrity of a system.

Please monitor the Redis Project mailing listsExternal Link for updates to ensure that you are aware of all updates to the software and can upgrade appropriately or apply patches and recompile as needed.

When upstream sources offer new releases, repeat the instructions for installing Redis and recompile your software when needed. These practices are crucial for the ongoing security and functioning of your system.

Managing Redis Instances

Running a Redis Datastore

In the default configuration, Redis runs in an interactive mode after being invoked at the command line. To start Redis in this manner issue the following command:

/opt/redis/redis-server /opt/redis/redis.conf.default 

You may now interact with Redis using any of the language specific bindings or use the built-in command line interface to interact with the Redis instance. Simply prefix any Redis commandExternal Link with the following string:


While running the Redis instance in this configuration is useful for testing and initial deployment, production deployments may have better results by creating a dedicated and unprivileged system user for the Redis instance and controlling Redis using an "init" script. This section covers the creation of an init script and strategies for managing production Redis instances.

Deploy Init Script

Issue the following sequence of commands to download a basic init script, create a dedicated system user, mark this file as executable, and ensure that the Redis process will start following the next boot cycle:

cd /opt/ wget adduser --system --no-create-home --disabled-login --disabled-password --group redis mv /opt/ /etc/init.d/redis chmod +x /etc/init.d/redis chown -R redis:redis /opt/redis touch /var/log/redis.log chown redis:redis /var/log/redis.log update-rc.d -f redis defaults 

Redis will now start following the next boot process. You may now use the following commands to start and stop the Redis instance:

/etc/init.d/redis start /etc/init.d/redis stop 

Managing Datastore Persistence

Redis is not necessarily intended to provide a completely consistent and fault tolerant data storage layer, and in the default configuration there are some conditions that may cause your data store to lose up to 60 seconds of the most recent data. Make sure you understand the risks associated and the potential impact that this kind of data loss may have on your application before deploying Redis.

If persistence is a major issue for your application, it is possible to use Redis in a transaction journaling mode that provides greater data resilience at the expense of some performance.

To use this mode, ensure that the following values are set in redis.conf:

File excerpt: redis.conf

appendonly yes appendfsync everysec 

The first directive enables the journaled "append only file" mode, while the second directive forces Redis to write the journal to the disk every second. Theappendfsync directive also accepts the argument always to force writes after every operation which provides maximum durability, or never which allows the operating system to control when data is written to disk which is less reliable.

After applying these configuration changes, restart Redis. All modifications to the data store will be logged. Every time Redis restarts it will "replay" all transactions and ensure that your data store matches the log. However, in some conditions this may render the data-store unusable while the database restores. To avoid this condition, issue the following command to the Redis command line:

/opt/redis/redis-cli bgrewriteaof 

You may wish to issue this command regularly, perhaps in a cron job, to ensure that the transaction journal doesn't expand exponentially. bgrewriteaofis non-destructive and can fail gracefully.

Distributed Data Stores with Master Slave Replication

Redis contains limited support for master-slave replication which allows you to create a second database that provides a direct real time copy of the data collection on a second system. In addition to providing a "hot spare" or multiple spares for your Redis instance, master-slave systems also allow you to distribute load amongst a group of servers. As long as all write options are applied to the master node, read operations can be distributed to as many slave nodes as required.

To configure master-slave operation, ensure that the following configuration options are applied to the slave instance:

File excerpt: redis.conf

slaveof 6379 

The slaveof directive takes two arguments: the first is the IP address of the master node, and the second is the Redis port specified in the master's configuration.

When you restart the slave Redis instance, it will attempt to synchronize its data set to the master, and then propagate the changes. Slave Redis instances can accept slave connections, which allows administrators to distribute the slave-replication load in multi-slave architectures. It's also possible to use a master with less stringent data persistence policies with a slave that keeps a more persistent copy. Master/slave replication creates a number of powerful architectural possibilities that may suit the needs of your application.

The traffic between slave instances and the master instance is not encrypted and does not require authentication in the default configuration. Authentication is available, and can be configured according to the documentation in the /opt/redis/redis.conf.default file; however, this is not the default method for securing Redis instances.

The preferred method for controlling access to Redis instances involves using iptables and possibly some sort of encryption such as an SSH tunnel to ensure that traffic is secure. Slaves will automatically attempt to reestablish a connection to the master node if the link fails in a number of situations. However, clusters cannot automatically promote members from slave status to master status; cluster management of this order must occur within your application.