Fleet
Upgrades
Upgrading gets the newest version of the manager onto your nodes. nginx keeps running throughout, so your sites stay up.
Doing it
cd /data/docker/failoverlb && sudo git pull
sudo /data/docker/failoverlb/install.sh --upgradeThat rebuilds the manager container, updates the host agent if it has changed, and starts everything again. It takes a couple of minutes.
The container carries the manager and the host carries the agent, so rebuilding the image on its
own only updates half of it. That is why the upgrade option exists rather than just a
docker compose up.
Do every node
Minutes apart, not days
A pair where the two are running different versions is a pair that can disagree about what it is able to do. While they differ, a new setting that only one of them understands is being shared between them. Do them one after the other, not at the same time, and check the Cluster page in between.
What is left alone
- Your environment file, including the secret key and the allow list.
- Your database: sites, pools, certificates, accounts, settings, everything.
- Your nginx build. That is upgraded separately from the build page.
Database schema changes are applied automatically at start up, and there is a check in the test suite specifically for a database made by an older version upgrading cleanly and keeping its rows.
Checking it worked
- Open the Cluster page. Both nodes online, both showing the same version.
- Apply your configuration once, even though nothing changed, to confirm both nodes still accept it.
That second step takes ten seconds and turns "it seems fine" into "it works".
Take a backup first
Not because upgrades usually go wrong, but because a backup taken before a change is the cheapest way to undo it, and the moment you want one is never the moment you have one.
Common questions
Is there downtime?
For the management screen, a couple of minutes per node. For your traffic, none. nginx is not restarted by an upgrade.
Can I skip versions?
Yes. Schema migrations run in order at start up.
What if the upgrade fails halfway?
The old container is still on the node. Fix whatever the output complained about and run the upgrade again. It is safe to run repeatedly.
Step by step instructions
The how to section has searchable, task shaped answers. Search it for upgrade.
Related features
The nginx build
nginx from source with 80 modules, rebuilt from the GUI.
Read moreBackup and restore
One encrypted file that rebuilds the fleet somewhere else.
Read moreClustering
Every node holds the whole config. No central controller.
Read moreChange history and rollback
Every apply kept, and put back with one click.
Read more