Install Redis on Fedora 36 – Quick Guide
If you’re running Fedora 36 and need a lightweight, in‑memory key/value store for caching or simple pub/sub work, this will get Redis up and running without pulling your hair out. We’ll use the distro’s package manager, enable the service, and do a quick sanity check with redis-cli.
Use the official packages – don’t reinvent the wheel
Fedora ships a well‑tested Redis build that matches the current stable upstream version. Compiling it yourself just to avoid “extra steps” usually ends up with missing SELinux policies or mismatched library versions. Trust the repo unless you have a very specific patch requirement.
Pull and install the package
sudo dnf install redis
dnf resolves dependencies, pulls the correct binaries for your architecture, and sets up the systemd unit file. Skipping this step and trying to copy a binary from somewhere else will likely break SELinux labeling and cause silent failures later.
Enable the service so it survives reboots
sudo systemctl enable --now redis.service
enable writes a symlink that tells systemd to start Redis on boot, while --now launches it immediately. If you only run systemctl start redis, the daemon disappears after the next reboot – not exactly what you want for a cache that should survive a power cycle.
Verify everything is alive
redis-cli ping
You should see PONG. That response means the server accepted a TCP connection, parsed the command, and replied correctly. If you get a timeout or “Connection refused,” double‑check the service status:
systemctl status redis.service
Look for any SELinux denials in journalctl -xe; Fedora is picky about unlabelled sockets.
Tweak the config (optional but handy)
The default /etc/redis.conf disables persistence, which is fine for a pure cache. If you need data to survive restarts, uncomment:
save 900 1
save 300 10
save 60 10000
Those lines tell Redis to snapshot the DB after a certain number of writes within given time windows. Remember to restart the service after editing:
sudo systemctl restart redis.service
A real‑world hiccup I’ve seen
After a routine Fedora 36 update, my colleague’s Redis stopped accepting connections. The culprit was a new SELinux policy that blocked the default Unix socket location (/var/run/redis/redis.sock). He fixed it by adding AllowTcpPort 6379 to his local policy module and reloading SELinux. Bottom line: always glance at /var/log/audit/audit.log when something “just stopped working” after an upgrade.
That’s it – Redis is now listening on port 6379, ready to serve as a cache or lightweight message broker. If you run into odd permission errors, check SELinux first; it loves to bite when you least expect it.
