Topology
Topology draws the servers in your cluster and the paths configuration travels between them, arranged so you can see at a glance which machines are up to date and which are behind.
It is one of the two views of the Servers screen, not a screen of its own: open Servers and use the selector at the top left to switch between Diagram and List. The view you choose is remembered.
Why you would want it
A list of servers does not show you relationships. It does not tell you that a call-processing server gets its configuration from the telephony frontend rather than from the manager, so when that server falls behind, the list gives you no hint about where to look. The diagram shows the path, so a server that is behind points you at the server that feeds it.
The list view remains the better one for anything you need to compare across servers - configuration versions, how far behind each server is, or how much each one carries - because those are columns and a diagram has no place to put them.
How it works
Each box is one server, labelled with what it does and the name you gave it. Each line is a configuration path, labelled with whether the server at the far end is up to date.
Configuration always starts at the manager and travels outward in two steps:
- The manager to the User Access server and the telephony frontend. These take configuration directly from the manager.
- The telephony frontend to the telephony workers. Workers take configuration from the frontend, not from the manager. In a cluster where the manager fronts telephony itself, the workers are drawn from the manager because that is where their configuration comes from.
The line colour and label follow the same states as Data Sync: a solid grey line means the server has applied the current configuration, and a dotted amber line means it has not yet.
Server types in the diagram
- Manager - runs the administration interface and is the origin of all configuration.
- User Access server - serves the user portal, Thirdlane Connect and the REST API.
- Telephony frontend - fronts call traffic and passes configuration on to the workers.
- Telephony worker - processes calls. A cluster can have as many as it needs.
An all-in-one system shows a single box, because one machine does all four jobs.
Using the diagram
Double-click any server to open its settings.
The toolbar carries one add button per server type the cluster can still take, and the type is filled in for you, so a new server lands in the right place in the diagram. A cluster holds only one User Access server and only one telephony frontend, so those buttons are shown greyed out once the server exists - the button stays visible to make it clear the server is already defined rather than the option being unavailable. A compact cluster shows only + Telephony worker, because the manager carries user access and telephony itself.
If a worker is drawn from the manager in a cluster that is supposed to have a separate telephony frontend, the frontend has not been defined. Cluster Configuration reports that as an error and refuses to publish, which is where to go to confirm it.
Adding or editing a server here changes a row in the Manager, not the cluster. When changes are waiting to be published, a notice appears above the diagram and takes you to the screen that publishes them.
Best practices
- Check the diagram after adding a server. A new server that appears in the wrong column has the wrong type set, and the type cannot be changed afterwards.
- Follow a dotted line back towards the manager. If several servers are behind at once, the one closest to the manager is usually the cause.
- Read it before a maintenance window. Knowing which servers take configuration from which tells you the order things will restart in.
Related documentation
- Servers - defining the servers the diagram draws.
- Cluster Configuration - publishing the cluster description to every server.
- Data Sync - per-server sync status and version detail.
- Deployment Options - cluster architecture and when to scale out.