Skip to content

Installation

There are multiple installation options for Karabo. A full system will require platform-compiled code and is currently only supported on x64 linux flavours with a reasonably modern (2.28+) glibc

Standalone and Full

If you would like to try Karabo in a standalone fashion on a single machine it's easiest to just run:

pip install karabo[full]

This will install a complete setup, including all essential service devices. Note that this requires, compiled code to be installed. This will be done for wheels on supported architectures, and a source compilate will be attempted otherwise.

If that fails, we recommend you follow the instructions at https://github.com/European-XFEL/Karabo which will result in a self-consistent environment, including a Python interpreter, with similar functionality to a karabo[full] installation.

If you require the GUI, you will need to install it separately into a different Pythonenvironment, as it does not require it's dependencies to align outside those of karabo.native:

pip install karabo.gui

After installation you should initialize the environment with either

karabo-activate --init-to PATH --standalone

for a standalone system, or 

karabo-activate --init-to PATH [--backbone] --broker-host amqp://{broker:port} --broker-topic {TOPIC}

for headless full installations. The --backbone option sets the core services up, and the --broker-* parameters are used to configure the installation according to your AMQP broker setup.

Additional options

  • karabo[karathon] will install the Karathon Python API
  • karabo[cpp] will install the Karabo C++ API
  • karabo[bound] will install the Karabo Python bindings to C++
  • karabo[services] will install service tools, allowing to manage installations with karabo.daemontools

Options can be combined, such as karabo[karathon,services], which will result in a Karathon installation with services tools installed.

To install the Karabo GUI, please use

pip install karabo.gui

Starting a Karabo standalone system

If you wish to run Karabo standalone you can do this if you have podman or docker installed on the system.

First, run

karabo-activate --init-to $PATH --standalone

Then run 

podman-compose -f $PATH/var/containers/compose.yaml up

or (docker-compose ...). This will start containerized version of the services a full Karabo installation requires:

  • A RabbitMQ broker (user: xfel, pw: karabo)
  • An Influx database instance for logging (user: infadm, pw: admpwd)
  • An ExistDB database as project database (user: admin, pw: change_me_please)
  • A Grafana installation with the Influx database as a pre-provisioned source (user: admin, pw: admin)

Note that the Grafana provisioning assumes karabo as the broker topic in use. If you've used karabo-activate --broker-topic TOPIC you'll need to edit the data source to refer to a database of TOPIC from inside Grafana.

Finally, run

$PATH/activate

followed by 

karabo-start

Device Development

These packages are not intended to facilitate development of devices for Karabo's C++ API. Please create a self-consistent development environment using the instructions at https://github.com/European-XFEL/Karabo.

Development of devices against these packages using the Bound and Karathon Python APIs should be trivially possibly, as long as the device packages define the correct entrypoints.

For the Karathon API, a line similar to this one should exist in your pyproject.toml:

[project.entry-points."karabo.middlelayer_device"]
PropertyTestMDL = "karabo.middlelayer_devices.property_test:PropertyTestMDL"

For the Bound API, a line similar to this one should exist in your pyproject.toml.

[project.entry-points."karabo.bound_device"]
PropertyTest = "karabo.bound_devices.property_test:PropertyTest"