Skip to main content

remote jupyter kernels via ssh tunnels

Project description

The design of this pakage is based upon SSH Kernel which in turn is based upon remote_ikernel. This implementation shares a common set of command line parameters with SSH Kernel, but it was implemented from scratch to adapt to recent changes to jupyter_client (which broke ssh_ipykernel) and to support Python 3.10. This package adds an ls implementation which allows listing info about the available kernel specifications.

While there are modest additions to SSH Kernel, there are also modest subtractions. There are fewer configuration options for things like the internal name used by Jupyter Client to refer to the created kernel.

Listing the Jupyter Kernels that are available

It can be difficult to know which Jupyter Kernels are available because there are multiple locations where the Kernel Spec files can be found. sshpyk has an ls option which lists the kernels that are available (even those which are not SSH Kernel or sshpyk kernels:

bash$
bash$ python -m sshpyk.kernel.ls --no-check -a
python3                                   /Users/drs/develop/python/conda/envs/py310/share/jupyter/kernels/python3
python3.8                                 /usr/local/share/jupyter/kernels/python3.8
python3dbg                                /Users/drs/Library/Jupyter/kernels/python3dbg
ssh__sshhost06test001                     /Users/drs/Library/Jupyter/kernels/ssh__sshhost06test001
ssh__sshhost06test002                     /Users/drs/Library/Jupyter/kernels/ssh__sshhost06test002
ssh_host06_host06homehost06condaenvspy310 /Users/drs/Library/Jupyter/kernels/ssh_host06_host06homehost06condaenvspy310
bash$

The --no-check (or alternatively -nc) flag indicates that the validity of the kernel spec files should not be checked. The -a (or --all) flag indicates that it should show all kernel specifications rather than just the ones for SSH Kernel or sshpyk kernel specification files.

If --no-check is not supplied, part of listing the kernel information will include verify that the Python executable specified in the kernel specification exist on the local and remote systems. This check allows the ouput to be colorized so red text indicates a problem. --local will limit the check to just the local Python executable and --remote will limit the check to only the remote Python executable. These options also list the local or remote Python path instead of the path to the kernel specification directory.

Command line “ls” options

The following options are available for listing the Jupyter kernel specifications:

–help, -h

Show help information and exit.

–all, -a

List all kernels that are available rather than just SSH Kernel and sshpyk kernels.

–local, -l

Only list the information for the local Python executable.

–remote, -r

Only list the information for the remote Python executable.

–no-check, -nc

Do not check for the existence local or remote Python executables. This option can be used alone or with other options, e.g. with -l. --no-check avoids colorization to indicate problems so it can be useful for scripting.

Adding a new Kernel for a Remote System

sshpyk can also be used to add a specification file for a remote Python Kernel. For a remote kernel to work

  • ssypyk must be installed on the local system

  • ipykernel must be installed on the remote system

Once these requirements are satisfied, the new kernel can be added like:

bash$ python3 -m sshpyk.kernel.add --host host06 --python /home/host06/conda/envs/py310 --display-name 'host06 kernel'

This will add a Python kernel which will run on host06, and it will be called host06 kernel when it is listed as an option for the user. We can check to see if it is now included by using the ls functionality:

bash$ python3 -m sshpyk.kernel.ls -r -a -nc
python3                                   localhost:python
python3.8                                 localhost:/opt/local/bin/python3.8
python3dbg                                localhost:/Users/drs/develop/casagui-ic-debugging/iclean_demo_venv/bin/python
ssh__sshhost06test001                     host06:/home/host06/conda/envs/py310/bin/python
ssh__sshhost06test002                     host06:/home/host06/conda/envs/py310/bin/python
ssh_host06_host06homehost06condaenvspy310 host06:/home/host06/conda/envs/py310/bin/python
ssh_host06_host06kernel                   host06:/home/host06/conda/envs/py310/bin/python
bash$

Unlike the example above, here we have asked that the remote Python path be displayed instead of showing the kernel specificaton directory. Since we again asked that all kernels be displayed instead of only the SSH kernels, a Python path is displayed for the non-SSH kernels, but it is the local Python path as indicated by localhost:. Because these three kernels are non-SSH kernels this is the only Python path that is available. However for the SSH kernels, we can see the remote Python path listed. These paths are prefixed with the hostname, here host06:. We can also see the newly added kernel listed as ssh_host06_host06kernel. This name is an internal name created from the --display-name string which the end user will typically never see.

Command line “add” options

–help, -h

Show help information and exit.

–timeout TIMEOUT, -t TIMEOUT

Specify timeout to wait for kernel startup text. This option is not used by sshpyk. It is only used by ssh_ipykernel.

–env [ENV …], -e [ENV …]

Specify environment variables for access by code executed within the remote Python kernel the form: "NAME=VALUE"

–display-name DISPLAY_NAME, -d DISPLAY_NAME

Specify the string to be used to describe this kernel to the end user.

–session

Signal that session information should be stored in ~/.sshpyk/sessions for this kernel”. This option is not used by ssh_ipykernel. Note that when --session is used the terminal output generated from the SSH connection is also directed to ~/.sshpyk/sessions.

–sudo, -s

Use sudo to start kernel on the remote machine. This option is not currently used by sshpyk. It is only used by ssh_ipykernel.

–host HOST, -H HOST

The name of remote host as used to connect with SSH.

–python PATH, -p PATH

Path to remote python installation. This is the path to the root of the Python installation so the Python executable would be found in <PATH>/bin/python.

SSH configuration notes

The host name used above is different from the Internet Protocol name for hosts which have a well defined address. It is also a name apart from the physical network where the host can be found. SSH allows for rationalizing the naming of the hosts to which you have access. This is done through the $HOME/.ssh/config file.

The most useful configuration option with respect to sshpyk is the ability to set up access to a host behind a bastion host. Assuming, the host named host06 from above is behind a bastion host, a configuration entry in $HOME/.ssh/config like:

Host host06
  User HOST06-USERNAME
  ForwardX11Trusted yes
  ProxyCommand ssh BASTION-USERNAME@ssh.example.com -W %h:%p

will allow the local account to use its SSH credentials for the user name BASTION-USERNAME on ssh.example.com for access to host06 which is on some protected network behind ssh.example.com. When the local account runs ssh host06, SSH will first connect as BASTION-USERNAME on ssh.example.com and then it connect to host host06 as username HOST06-USERNAME.

This sort of configuration will allow host06 to be allowed as a hostname for remote kernels.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

sshpyk-0.16.tar.gz (12.8 kB view details)

Uploaded Source

Built Distribution

sshpyk-0.16-py3-none-any.whl (12.7 kB view details)

Uploaded Python 3

File details

Details for the file sshpyk-0.16.tar.gz.

File metadata

  • Download URL: sshpyk-0.16.tar.gz
  • Upload date:
  • Size: 12.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/5.0.0 CPython/3.12.3

File hashes

Hashes for sshpyk-0.16.tar.gz
Algorithm Hash digest
SHA256 61812341b64dc05b96514a8b1d6a143e7ffc7d72ecf4ba46af7160b3bf570917
MD5 2eea18c43dfea7f150a95a2b01fd972e
BLAKE2b-256 bd438d32c9e8c60dc35f1f37e85d99c23d2d8da9253d0ffe9e492b94fa802724

See more details on using hashes here.

File details

Details for the file sshpyk-0.16-py3-none-any.whl.

File metadata

  • Download URL: sshpyk-0.16-py3-none-any.whl
  • Upload date:
  • Size: 12.7 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/5.0.0 CPython/3.12.3

File hashes

Hashes for sshpyk-0.16-py3-none-any.whl
Algorithm Hash digest
SHA256 96f2c69400d5a4339db9f36899f2b80f5186e39e900d709b0264eea57ac4c91c
MD5 a770212e850860d69cc98abf8ac5a2fd
BLAKE2b-256 5733f53294195cf40715ff2eee763e88388edecb260785cfce841da4c64891ab

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page