Open cisaacstern opened 1 year ago
Note that I don't think we should try flink on HPC, described as more traditional slurm-like setup.
I remember @negin513 mentioning they may have a kubernetes cluster - so this should be about trying flink on an on prem kubernetes cluster, rather than any HPC system. The on-prem cluster may be run by a HPC center though :)
Important clarification, thanks @yuvipanda!
With #114, we should be able to run Pangeo Forge on Flink on HPC. This may not be quite as simple as an eventual Dask deployment, but #109 is realistically a longer-term project, so Flink is likely our best option for now.
Tagging @negin513 and @TomNicholas as possibly interested in trying this out at NCAR. If either of you do have interest in this, #124 should be helpful once it goes in.