Software architecture at Sirius’ MOGNO beamline, towards a web-based system

Abstract MOGNO beamline, the micro and nanotomography beamline of the synchrotron light source Sirius, at the Brazilian Synchrotron Light Laboratory (LNLS), supports a variety of tomographic experiments including regular, zoom and panoramic tomographies, and high throughput, as well as in-situ experiments. The existing control software is based on EPICS and uses Python and the PyEpics library for the device abstraction layer that is made available for different control routines. To ensure ease of use, these routines are maily accesed via PyQt/PyDM GUIs at the beamline’s control terminals. Looking to reduce the dependency on the control terminals, a transition towards a web-based control system is proposed, where maintainability, availability and safety, are enhanced. The new architecture is based on a microservices that provide APIs for experiment control and data processing, centralizing tasks on a server and allowing for access through the same GUIs already in use with few changes. Moreover, the decoupling will enable the move to a standardized backend, following facility-wide efforts, such as the use of the Bluesky stack. Importantly, the proposed architecture also enables the use of standard tools for security on its different layers, such as authentication methods, HTTPS, and user management. As proof of concept, the system has been deployed for data reconstruction at the beamline, managing job submission to an HPC cluster via a PyQt GUI.