Connect Spyder IDE to a remote kernel on Mistral#

I am just describing spontaneously what worked for me to connect my local Spyder instance to a remote node on Mistral THAT YOU CAN CONNECT TO VIA SSH FROM YOUR LOCAL MACHINE!!!!

This is just a draft tutorial that will be updated/optimized afterwards.

Warning

  • If you have ssh issues on your side, this post is not for you

  • Some users experienced issues on Mac and Windows. Unfortunately, these are client-specific and not related to Mistral

  • If you have a workaround, you can share it with other users

Requirements#

  • did you read Spyder documentation?

  • did you follow the steps described here?

  • do you know that you have to create the json configuration file?

  • …

This is what worked for me#

Note

I tried these steps on my local Ubuntu 20 laptop and mistralpp4 (random) as the remote machine. It can work on other nodes (partitions) but I didn’t test it yet. Please, note that most of the issue are on the user side, for Spyder-specific issues check their [troubleshooting guide](https://docs.spyder-ide.org/current/first-steps.html).

Log-in (ssh) to mistralpp and get ip address#

  1. ssh <username>@mistralpp.dkrz.de

  2. curl ifconfig.me

you need the ip addr!

Activate your conda env and start the kernel#

Note

Better to have spyder installed (spyder-kernels) on your conda env on Mistral. Check their documentation.

  1. module load python3/unstable

  2. source activate <your env>

  3. python -m spyder_kernels.console --ip=<ip_addr_of_the_node>

This will generate a kernel-xxxx.json in .local/share/jupyter/runtime/ THAT YOU NEED TO COPY IT TO YOUR LOCAL MACHINE!!! with scp for example.

Configure your local Spyder instance#

This step is similar to the one described here? These are the parameters:

image0

That’s all!

Troubleshootings#

  • I assume that the Spyder remote concept is based on this. In case you are not able to connect to the remote kernel, you can use that tutorial for debugging.

References#

[1] How to connect your Spyder IDE to an external ipython kernel with SSH PuTTY tunnel

[2] Connecting Spyder IDE to a remote IPython kernel

[3] Official Spyder IDE step-by-step