I would be grateful if users could share successful cases of CryoSPARC installation and usage on GPU clusters. If you have deployed CryoSparc on your local cluster, could you share your experiences? How many active users do you have at the moment? How do they access data for processing? How the access to the web site is configured? We test Cryosparc on our GPU cluster here at the ETH Zurich and this info would be very useful for the future configuration and usage of the program.
I am running a cryosparc instance on a Linux cluster. It has a slurm based schedular. There is a shared nfs home directory and lustre filesystem. I have installed the client code on a directory on the NFS filesystem, as well as the database. The server code runs in a Singularity container. Another Singularity container runs a nginx proxy server that faces the network, providing virtual hosts and ssl termination. There is a subset of the compute nodes that have 2 GPU’s installed in them and a 1 TB SSD drive. Slurm jobs schedule on these nodes.