How To Connect to a Managed Redis Instance over TLS with Stunnel and redis-cli

Introduction

A managed Redis instance can provide benefits like high availability and automated updates. However, any time you make a connection to a remote database server, you run the risk of malicious actors sniffing the sensitive information you send to it.

redis-cli, the Redis command line interface, doesn’t natively support connections over TLS, a cryptographic protocol that allows for secure communications over a network. This means that without further configuration, redis-cli is not a secure way to connect to a remote Redis server. One way to establish a secure connection to a managed Redis instance is to create a tunnel that uses the TLS protocol.

Stunnel is an open-source proxy used to create secure tunnels, allowing you to communicate with other machines over TLS. In this guide, we will walk through installing and configuring stunnel so you can connect to a managed Redis instance over TLS with redis-cli.

Prerequisites

To complete this guide, you will need:

• Access to an Ubuntu 18.04 server. This server should have a non-root user with administrative privileges and a firewall configured with ufw. To set this up, follow our initial server setup guide for Ubuntu 18.04.
• A managed Redis database instance. The steps outlined in this tutorial were tested on a DigitalOcean Managed Redis Database, though they should generally work for managed databases from any cloud provider. To provision a DigitalOcean Managed Redis Database, follow our Managed Redis product documentation.

Step 1 — Installing Stunnel and redis-cli

When you install a Redis server, it usually comes packaged with redis-cli. However, you can install redis-cli without the Redis server by installing the redis-tools package from the default Ubuntu repositories. You can also install stunnel from the default Ubuntu repositories by downloading the stunnel4 package.

First, update your server’s package index if you’ve not done so recently:

sudo apt update

Then install the redis-tools and stunnel4 packages with APT:

sudo apt install redis-tools stunnel4

When prompted, press ENTER to confirm that you want to install the packages.

You can check whether stunnel was installed correctly and its systemd service is working by running the following command:

sudo systemctl status stunnel4

Output
● stunnel4.service - LSB: Start or stop stunnel 4.x (TLS tunnel for network daemons)
   Loaded: loaded (/etc/init.d/stunnel4; generated)
   Active: active (exited) since Thu 2019-09-12 14:34:05 UTC; 8s ago
     Docs: man:systemd-sysv-generator(8)
    Tasks: 0 (limit: 2362)
   CGroup: /system.slice/stunnel4.service

Sep 12 14:34:05 stunnel systemd[1]: Starting LSB: Start or stop stunnel 4.x (TLS tunnel for network daemons)...
Sep 12 14:34:05 stunnel stunnel4[2034]: TLS tunnels disabled, see /etc/default/stunnel4
Sep 12 14:34:05 stunnel systemd[1]: Started LSB: Start or stop stunnel 4.x (TLS tunnel for network daemons).

Here, you can see that the stunnel service is active, though the process immediately exited. This tells us that stunnel is running, but it isn’t able to actually do anything since we haven’t yet configured it.

Step 2 — Configuring Stunnel

Modern Linux systems rely on systemd for initializing and managing services and daemons. However, stunnel uses a SysV-style init script, which is based on the older UNIX System V init system, for startup. You’ll need to modify the /etc/default/stunnel4 file to enable this init script.

Open this file with your preferred text editor. Here, we’ll use nano:

sudo nano /etc/default/stunnel4

Find the ENABLED option near the top of the file. It will be set to 0 by default, but change this to 1 to enable stunnel to start at boot:

# /etc/default/stunnel
# Julien LEMOINE <speedblue@debian.org>
# September 2003

# Change to one to enable stunnel automatic startup
ENABLED=1
. . .

Save and close the file. If you used nano to edit the file, do so by pressing CTRL+X, Y, then ENTER.

Next, you will need to create a configuration file for stunnel which will tell the program where it needs to route traffic.

Open a new file called stunnel.conf under the /etc/stunnel directory:

sudo nano /etc/stunnel/stunnel.conf

Add the following content to this file:

fips = no
setuid = nobody
setgid = nogroup
pid = /home/sammy/pids/stunnel.pid
debug = 7
delay = yes
[redis-cli]
  client = yes
  accept = 127.0.0.1:8000
  connect = managed_redis_hostname_or_ip:managed_redis_port

The first five lines in the file are global options, meaning they will apply to every service you include in this file:

• fips: Enables or disables stunnel’s FIPS 140-2 mode. In this mode, stunnel will validate that the connection meets the Federal Information Processing Standard. Setting this to no disables this feature. Note that disabling this is not any less secure, but keeping it enabled (as it is by default) would require some extra configuration.
• setuid: Defines the Unix user ID under which stunnel will run. By default, the stunnel process is owned by the root user. However, the stunnel documentation recommends that you drop administrative privileges once the tunnel starts, as failing to do so poses a security risk. Setting the setuid parameter to nobody will cause nobody, an unprivileged user, to take ownership over the stunnel process once the tunnel has been established.
• setgid: Defines the Unix group ID under which stunnel will run. As with setuid, this configuration specifies a group without any special privileges — nogroup — to avoid any potential security issues.
• pid: Defines a file location where stunnel will create a .pid file, a type of file that contains a process’s PID. .pid files are typically used by other programs to find the PID of a running process. By default, stunnel creates a .pid file in the /var/run/stunnel4/ directory but because the nobody user doesn’t have permission to access that directory, it will prevent the tunnel from starting correctly. Instead, this line specifies a file named stunnel.pid held in a directory called pids in the Ubuntu user’s home directory. We will create this directory and the stunnel.pid file shortly. As you add this line, be sure to change sammy to the name of your Ubuntu system user.
• debug: Sets stunnel’s debugging level, which can range from 0 to 7. In this example we’ll set it to 7, the highest level available, as that will provide the most detailed information if stunnel runs into any issues. You can set it to any level you like, but be aware that the default setting is 5.
• delay: When set to yes, this option causes stunnel to delay the DNS lookup for the address listed in the connect option. It will also prevent stunnel from caching IP addresses. This setting will help to keep the tunnel open even if the managed Redis instance goes offline, as can happen when scaling up your cluster.

The remaining lines are service-level options, and only apply to the tunnel we’ll create for redis-cli:

• [redis-cli]: This is a service name and specifies that the following lines represent an individual service configuration for a client program. You can have more than one service in a stunnel configuration file, though each must be associated with an existing client application and you can’t have two services for the same application.
• client: Setting this to yes tells stunnel to run in client mode, meaning that stunnel will connect to a TLS server (the managed Redis instance) rather than act as a TLS server.
• accept: Defines the host and port on which stunnel will accept connections from the client. Here, we specify the IP address 127.0.0.1, which is an IPv4 loopback address used to represent localhost, and port 8000. This means stunnel will listen for connections originating from the Ubuntu server on port 8000 and encrypt them. Note that you can set the port to any port number you like as long as it’s not already in use.
• connect: Defines the remote address and port to which stunnel will make the connection. Be sure to change this parameter to align with your managed database’s port and hostname or IP address.

This is a fairly minimal configuration that leaves many of stunnel’s default settings in place. The program has many options available for you to create tunnels that suit your particular needs. See the official documentation for more details.

After adding this content, save and close the file.

Next, make sure you’re in your Ubuntu user’s home directory:

cd

Then create the pids directory which will hold the stunnel.pid file:

mkdir pids

The stunnel process will create the stunnel.pid file automatically, so you don’t need to create the file yourself. However, you do need to change ownership of the pids directory over to the nobody user and nogroup group:

sudo chown -R nobody:nogroup pids/

Then, restart the stunnel4 service so stunnel will read the new configuration file:

sudo systemctl restart stunnel4

Following that, you can test whether stunnel successfully created a tunnel by invoking netstat, a command line utility used to display network connections. Run the following command, which pipes the netstat output into a grep command, which in turn searches it for every instance of stunnel:

sudo netstat -plunt | grep stunnel

Output
tcp        0      0 127.0.0.1:8000         0.0.0.0:*               LISTEN      17868/stunnel 

This output shows that stunnel is listening for connections on local port 8000.

You can also confirm that the nobody user has taken ownership over the stunnel process with ps, a program that displays all currently-running processes:

ps aux | grep stunnel

Output
nobody   15674  0.0  0.1 121912  3180 ?        Ssl  19:28   0:00 /usr/bin/stunnel4 /etc/stunnel/stunnel.conf
. . .

Here, you can see that nobody has indeed taken over the stunnel process.

Stunnel is now fully configured and running on your system. You’re ready to connect to your managed Redis instance and test that the tunnel is working as expected.

Step 3 — Connecting To Your Managed Database over TLS

Now that you’ve installed redis-cli and configured stunnel on your server, you’re ready to connect to your managed database over TLS.

Based on the settings defined in the configuration file created in Step 2, you would connect to your managed database with the following command:

redis-cli -h localhost -p 8000

This command includes the -h flag, which tells redis-cli that the next argument will be the host to connect to. In this case, it’s localhost since we’re connecting to a tunnel created locally on the server. After that is the -p flag, which precedes the port of the local tunnel we’re connecting to, which in this case is port 8000.

After running that command, you will be connected to your managed Redis server. Your prompt will change to reflect that you’ve connected and are in redis-cli’s interactive mode:

redis-cli -h localhost -p 8000 -a password

auth password

You can test whether the tunnel is working as expected by running the ping command from Redis’s interactive mode:

ping

If the connection is alive it will return PONG:

Output
PONG

If, however, stunnel is not tunnelling traffic from your server to your Redis instance correctly, you may see an error message like this before being disconnected from Redis:

Output
Error: Server closed the connection

If you receive this or a similar error, double check that you’ve entered your Redis instance’s hostname and port correctly in your stunnel.conf file. Likewise, be sure that you entered the correct port number in your redis-cli command.

You may also receive this error when the managed Redis instance resets its open connections, as can happen when you scale up your cluster’s configuration. In such cases, you may not be disconnected from Redis, although you will need to reauthenticate.

Once you’ve confirmed that the tunnel is working, go ahead and disconnect from your Redis instance:

exit

If you ever change stunnel’s configuration, you’ll need to reload or restart the stunnel4 service so stunnel will notice the changes:

sudo systemctl reload stunnel4

If at any point in the future you want to close the TLS tunnel, you can also do so with systemctl:

sudo systemctl stop stunnel4

After the tunnel has ben closed, you can reopen the tunnel by starting the service again:

sudo systemctl start stunnel4

Now that you’ve successfully configured stunnel, you’re all set to begin adding data to your managed Redis instance with redis-cli.

Conclusion

Stunnel is a handy tool for creating TLS tunnels and establishing secure connections to remote servers. This is especially useful in cases where the secure transport of information between machines is critical, as with a remote database.

From here, you can begin exploring Redis and integrating it with your next application. If you’re new to working with Redis, you may find our series on How To Manage a Redis Database useful.