deepmodeling/dflow

remote slurm connection

xuhan-96 opened this issue · 2 comments

xuhan-96 commented

Hello, I am trying to connect slurm on my local machine and submit jobs remotely.

My remote slurm function is:
slurm_remote_executor = SlurmRemoteExecutor( host="**.**.**.**", port=22, username="**", private_key_file="$HOME/.ssh/id_rsa", workdir="***", header="***", )

When I execute this function, the logging information is: Permission denied (gssapi-with-mic, password)

So how to solve this problem? Do I need to make other configurations on my local machine? So far I only generated RSA key and put the public key on slurm system ($HOME/.ssh/authorized_keys).

By the way, normally when I ssh the slurm, I need double authoriation, both password/public key and OTP.

zjgemi commented

Double authorization usually means manual intervention by design, which is not likely to be done automatically. Maybe you can try if DPDispatcher supports for this. But I don't think so.

xuhan-96 commented

Double authorization usually means manual intervention by design, which is not likely to be done automatically. Maybe you can try if DPDispatcher supports for this. But I don't think so.

Thanks for your reply, I will check that!