Secure write for connection file #469
Conversation
|
I did some python 2 / 3 testing with downstream libraries as well and didn't run into any issues. |
|
@minrk @Carreau @SylvainCorlay The same change is now merged in jupyter-server/jupyter_server#77 if one of you could help review |
|
It seems the added dependency from this PR did not make it into the conda forge package, jupyterlab/jupyterlab#7259 . I'll create an issue on https://github.com/conda-forge/jupyter_client-feedstock to track this. |
|
This change to jupyter_client prevents running on the server? See https://stackoverflow.com/questions/58214911/kernel-error-jupyter-client-connect-py-assertionerror I am running into this issue in jupyterhub. |
|
It might be related to conda-forge/jupyter_client-feedstock#35? The conda recipes are their own bag of abstractions wrapping the source code so it can be difficult to untangle what's a source issue and what's build issue therein. Can you post the |
|
Here is the output from conda list in the Jupyter project datascience-notebook built December 2, 2019 03:31 PM. This is running in an Azure AKS cluster with the following storage permissions: `kind: StorageClass
I confirmed I get the error: File "/opt/conda/lib/python3.7/site-packages/jupyter_client/connect.py", line 105, in secure_write image: jupyter/datascience-notebook:latest Also check jupyter/notebook#4937 `jovyan@jupyter-shamberger-20michael-20-28ext-29:~$ conda list packages in environment at /opt/conda:Name Version Build Channel_libgcc_mutex 0.1 main conda-forge |
|
Issue is related to azurefile CIFS filesystem. Jupyter core has been updated in 4.6.1 to allow execute on owner. I am now trying to change group and world kubernetes pod definition built from kubespawner in jupyterhub. |
|
Because you're using I believe @MSeal is right about there being a conda-forge issue and it looks like @bollwyvl is on the case - thank you! |
- Recent addition in jupyter_client caused this bug. This should be removed later, once a newer version of jupyter_core is released. The offending code was added to jupyter_client, moved to jupyter_core, then fixed in jupyter_core. - original patch: jupyter/jupyter_client#469 (2019-09-08) - fix in jupyter_core: jupyter/jupyter_core#177 (2019-11-16)
Includes release information, as we should follow the merge directly with a patch release for 5.3.2.