aisphere/ffdl-cli

CLI for Fabric for Deep Learning (FfDL)

ffdl-cli's Issues

Will freeze if FFDL_GRPC_CERT is set incorrectly

The CLI lacks proper error handling in case there is a certificate mismatch, i.e. if FFDL_GRPC_CERT is set incorrectly, ffdl list will just freeze until timeout. It should rather print an expressive error message saying that the certificates don't match.

Will freeze if VPN is not established

If the VPN connection was not properly established for DLaaS deployment, the CLI will just freeze rather than throw an appropriate error. I'm not sure how to reliably identify this case, but maybe we can at least print out a suggestion.

[I encountered the problem, since I had set my Proxifier rule to direct connection due to misrouting for other IBM IP things I was developing and it took me a while before I noticed.]