I am realizing that our current workload of developing on nvidia VMs is deviating significantly from how the stack run on k8s environment. So the shims or abstraction we introduced have increased the gap between dev and prod environments.
So the task here is to investigate and figure out dev/prod environment workflow that reduces the toil, maintains the faster dev loop.
Some ideas:
- using
kind for dev environments and explore mounting the files in the pods through local filesystem for faster dev cycles.
/cc @ShubyM
I am realizing that our current workload of developing on nvidia VMs is deviating significantly from how the stack run on k8s environment. So the shims or abstraction we introduced have increased the gap between dev and prod environments.
So the task here is to investigate and figure out dev/prod environment workflow that reduces the toil, maintains the faster dev loop.
Some ideas:
kindfor dev environments and explore mounting the files in the pods through local filesystem for faster dev cycles./cc @ShubyM