I am looking to make use of Open OnDemand to front the QuickEMU project. It would be great for our researchers to spin up ephemeral instances of various operating systems or applications which provide visualization capabilities or a suite of applications specific to a research domain within HPC. I am aware of on-premise cloud solutions which will provide this but Open OnDemand provides a one-click up-and-running interface ( Science Gateway ) approach to lower the barriers.
Just putting the idea out there or if anyone has done something similar with another tool, let me know.
Hi! Thanks for the topic. I’m not aware of anyone who’s implemented QuickEMU as an adapter (that’s the term we use to talk about different schedulers).
I know someone has done something with OpenStack, though to my knowledge it was never open sourced.
If you don’t have an Openstack cloud deploying it only for this use case can be overkill but maybe you can adapt the slurm config in your environment so the slurm master uses the QuickEMU api instead of the openstack api. In our setup the slurm master is using these two scripts to start/stop the machines using the Openstack api, you would only need to adapt those two script to use the quickemu api