Running a Brayns backend service

For any kind of vizualization, a running instance of Brayns renderer backend with a websocket server (braynsService) is required to perform the computationally expensive tasks.

An instance can be started using its executable (braynsService binary) using the terminal. A typical command line to start it is the following.

$ braynsService --uri 0.0.0.0:5000 --plugin braynsCircuitExplorer --plugin braynsAtlasExplorer

It will start a Brayns instance on port 5000 with two plugins to handle circuits and atlases.

Websocket server

A brayns instance runs a websocket server if an URI is specified in the command line and can be monitored from an external application using JSON-RPC messages.

The URI format is always <IP>:<PORT> where the IP specifies the addresses which are authorized to connect to the server and PORT the communication port used by the instance.

Here is an example of server running on port 5000 accepting connections from any host (wildcard IP).

$ braynsService --uri 0.0.0.0:5000

Plugins

Plugins can be loaded to perform more than the basic functionalities of Brayns using the command line.

Here is how to start brayns with the CircuitExplorer plugin:

$ braynsService --plugin braynsCircuitExplorer

Hint

Brayns, by default, does not understand science-specific models. These specific functionalities are loaded using plugins. The CircuitExplorer plugin is the main plugin for neuroscientific visualization that you will want to load.

For further information on plugins, check Brayns plugins.

Log level

The instance log level can be configured using –log-level followed by either trace, debug, info, warn, error, critical or off. This is useful for debugging and see the requests sent from and received by the instance.

$ braynsService --log-level debug

Help

The command line usage of braynsService can be displayed using:

$ braynsService --help

Running an instance on BB5

On BB5 the binary executable of braynsService is provided through spack and can be run with the following steps.

• 1 - Allocate a node on BB5:

$ salloc --account=<your_project_account> -p interactive -t 8:00:00 --exclusive --constraint=cpu -c 72 --mem 0

Where your_project_account should be the project account you have access to.

After a successful allocation, an SSH connection with the node is opened automatically. The node ID is composed of alpha-numeric characters, for example: r2i2n11.

• 2 - Load brayns module:

To access Brayns binary executable, user must load brayns module using:

$ module load unstable
$ module load brayns/latest

• 3 - Launch the braynsService backend instance:

Now a Brayns instance can be started as before:

$ braynsService --uri 0.0.0.0:5000 --plugin braynsCircuitExplorer

Now we know the host (r2i2n11) and port (5000 in this case) where our Brayns backend is running. We can connect to it through the Python client (here using uri=’r2i2n11:5000’).

Hint

If the websocket client (e.g. Python script) is running outside BB5, r2i2n11 should be replaced by r2i2n11.bbp.epfl.ch.

Now you can interact with the backend instance by using the Python API Using Brayns Python API.