Here I refactored the compute resource daemon in order to implement the scheme where we do not need to restart the compute resource for changes in the app configuration (from the frontend) to take effect.
Separated out the daemon into 3 classes: Daemon, AppManager, and JobManager
When the apps are changed in the frontend, a pubsub message is sent, and that's when the app manager knows to reload the apps.
TODO: reset the slurm job handler when an app is reloaded.
@luiztauffer
Here I refactored the compute resource daemon in order to implement the scheme where we do not need to restart the compute resource for changes in the app configuration (from the frontend) to take effect.
Separated out the daemon into 3 classes: Daemon, AppManager, and JobManager
When the apps are changed in the frontend, a pubsub message is sent, and that's when the app manager knows to reload the apps.
TODO: reset the slurm job handler when an app is reloaded.