Clusters: the master and worker nodes
What a cluster is in CreateYourVPN: the master (management node), worker nodes, the cluster page, changing the master, and the lifecycle — in plain words.
In the previous lesson we compared a cluster to an office complex: one administration office and the working buildings. Now let's take a closer look at a cluster — what it's made of, how it's created, and what you see on its page.
Why clusters at all
One server is simple: everyone's on it. But the moment you have two servers, questions appear: where do you keep the shared user database? How do the servers "know" about each other? How do you give someone access to all servers with a single link?
A cluster solves all of that: the servers join into one system with a shared user database and shared management. A user connects "to your VPN", not to individual machines.
The master (management node)
The master is the brain of the cluster. It:
- stores the cluster's user database: accounts, limits, expiry dates;
- issues certificates to the worker nodes — the "passes" nodes use to prove they belong;
- hands out configuration to the nodes: which inbounds to run and with what settings.
There is always exactly one master per cluster. On its own, it doesn't serve users — that's the worker nodes' job.
Worker nodes
The nodes are the hands of the cluster: your users' traffic flows through them. Nodes are where inbounds live — the entry points into the VPN. A single cluster can hold up to 100 nodes.
The master and a node can live on the same server. For your first server this is the default and the recommended setup: one machine both manages the cluster and serves VPN traffic. When you get more users — just add more nodes.
How a cluster is created
A cluster is born together with your first server: click "Connect a server", give the cluster a name (for example, EU · production), and enter the SSH credentials of your VPS — everything else happens automatically. What exactly happens at each stage of installation is covered in the next lesson.
New servers join an existing cluster via the "Add node" button: during installation, the node automatically receives a certificate from that cluster's master and connects to it.
A cluster has its own states, shown on its card: "Provisioning" (installation in progress), "Active" (everything's running), "Degraded" (something went wrong), "Archived" (the cluster was deleted).
The cluster page
Open a cluster and you'll see its whole anatomy, top to bottom:
- Topology diagram — a live map: the master (management node) on the left, then the worker nodes with their inbounds, and the routes on the right. You can drag inbounds between routes right here with the mouse.
- Routes — the cluster's list of routes (lesson 4).
- Servers — cards for every machine in the cluster with metrics: online users, traffic, CPU, RAM (lesson 3).
Click a server or an inbound to open the inspector with the details: address, state, load coefficient, internal identifiers.
Changing the master ("Make this the master")
Sometimes the master needs to move: you've found a more reliable server, or the current machine is having provider issues. That's what the master swap is for: in the inspector of any connected node, click "Make this the master" — the node becomes the master, and the current master turns into a regular node. The operation is reversible.
What's important to know:
- all users and settings migrate to the new master automatically;
- user subscriptions keep working — nobody has to reconfigure anything; though the two participating servers (the new and the old master) may be briefly unavailable;
- the process runs in the background and can take up to an hour; each node briefly reconnects to the new master;
- starting it requires a code from an email — protection against an accidental click;
- while the swap is running, changes to the cluster topology (nodes, inbounds, routes) are locked.
Deleting a cluster
A cluster is deleted together with its master server: the panel will warn you that this archives the entire cluster. The rented VPS itself keeps running — CreateYourVPN simply stops managing it. Deleting a worker node is more modest: only that node leaves the cluster (along with its inbounds — the panel shows exactly what the deletion will affect).
Remember
- A cluster = one master + up to 100 worker nodes, with a shared user database.
- The master manages, the nodes carry the traffic. Both can live on one machine.
- A cluster is created with your first server; nodes are added in a couple of clicks.
- The master can be "moved" to another node with a master swap — invisibly to your users.