req_keepvars implies that it is for setting environment variables - but to use it, you have to re-add all of the JH defaults. This is not very intuitive, and makes it not very useful.
Rename req_keepvars to req_keepvars_default - this is pre-filled by default with the output from jupyterhub.spawner.Spawner.get_env() which is the environment that JH thinks that singleuser servers need.
Add a new req_keepvars which gets added to req_keepvars_default to become the actual req_keepvars in the batch script. This and previous would probably have helped with #118,
Add a new admin_environment which is added to the environment for running the commands, but not the --export={keepvars} in the scripts. This can be used to authenticate to batch systems in something as an admin. Note that it is not yet guaranteed that this works in all spawners.
Any thoughts on these?
All together these help with #82, considering security.
req_keepvars
implies that it is for setting environment variables - but to use it, you have to re-add all of the JH defaults. This is not very intuitive, and makes it not very useful.Rename
req_keepvars
toreq_keepvars_default
- this is pre-filled by default with the output fromjupyterhub.spawner.Spawner.get_env()
which is the environment that JH thinks that singleuser servers need.Add a new
req_keepvars
which gets added toreq_keepvars_default
to become the actualreq_keepvars
in the batch script. This and previous would probably have helped with #118,Add a new
admin_environment
which is added to the environment for running the commands, but not the--export={keepvars}
in the scripts. This can be used to authenticate to batch systems in something as an admin. Note that it is not yet guaranteed that this works in all spawners.Any thoughts on these?
All together these help with #82, considering security.