Cannot connect kernel from VS Code

[tdevinda]

I am trying to use a notebook in my Fabric Workspace from VS Code. 

The issue I encounter is that my VS code does not connect to the pyspark kernel. I tried the methods highlighted here: (https://community.fabric.microsoft.com/t5/Fabric-platform/Assistance-Needed-Kernel-Connection-Timeou...)

Setting the timeout did not solve the issue and I waited more than 10minutes but my cell does not execute. 

I had on-and-off functionality a few weeks back but generally, the notebook worked in VS Code when I do Publish, and merge. It was a hassle but worked ok. This was for runtime version 1.2

Now it seems the online default runtime is 1.3. The new libs got installed on my conda environment when I ran it for the first time. But the code does not run.

I tried changing the notebook's runtime version to 1.2 in the web console but that did not do any changes as well. 

VS Code version is: 1.99.3

The monitor goes from not started -> In Progress when I run the cell. But the cell does not execute. 

Any known bugs/workarounds would be appreciated. 

[v-hashadapu]

Hi @tdevinda , Please let us know if your issue is solved. If it is, consider marking the answer that helped 'Accept as Solution', so others with similar queries can find it easily. If not, please share the details.
Thank you.

[v-hashadapu]

Hi @tdevinda , Please let us know if your issue is solved. If it is, consider marking the answer that helped 'Accept as Solution', so others with similar queries can find it easily. If not, please share the details.
Thank you.

[tdevinda]

Hi,

For the current kernel version 1.3, the issue was never resolved. I am waiting till some stable fix comes for 1.3. 

Even for a fresh install, the 1.3 kernel's notebooks try to find a conda environment in VS Code. I think the extension needs an update. 

[v-hashadapu · Accepted solution]

Hi @tdevinda , Thank you for reaching out to the Microsoft Community Forum.

This may be due to Fabric VS Code extension not being fully updated to support remote kernel binding for runtime 1.3. The extension still relies on patterns from runtime 1.2, where local execution was viable, causing the current incompatibility. This is not user error but a tooling limitation.

Currently, Microsoft has not yet released full support for remote notebook execution through VS Code for runtime 1.3. This limitation has been noted across the community, and an update to the extension is expected, but no timeline has been published. Until then, the Fabric web UI remains the only fully supported environment for executing notebooks using runtime 1.3, as it correctly provisions and binds the remote compute kernel.

If you want to monitor progress or confirm when support is added, the best place to watch is the Microsoft Fabric release notes .

If this helped solve the issue, please consider marking it 'Accept as Solution' so others with similar queries may find it more easily. If not, please share the details, always happy to help.
Thank you.

[v-hashadapu]

Hi @tdevinda , Please let us know if your issue is solved. If it is, consider marking the answer that helped 'Accept as Solution', so others with similar queries can find it easily. If not, please share the details.
Thank you.

[v-hashadapu]

Hi @tdevinda , Thank you for reaching out to the Microsoft Community Forum.

Runtime 1.3 uses a remote compute model, where execution happens on Fabric’s cloud infrastructure, not on your local Conda environment. Your previous success with runtime 1.2 was due to local execution, which is no longer compatible with 1.3. The kernel mismatch causes VS Code to stall when trying to execute using an unsupported environment.

First run your notebook directly in the Fabric web UI to confirm the kernel starts and runs correctly. If it does, the problem is isolated to VS Code’s kernel selection. In VS Code, select the Fabric runtime kernel in the dropdown in the top right corner of the notebook interface. This will link you to Fabric’s remote compute and resolve the stall. Avoid using local Conda environments for runtime 1.3 notebooks, these are unsupported in the current model.

If you want to continue working with a local execution setup, you must create a new notebook in the web UI, set its runtime to 1.2, then sync it to VS Code. You can then use a properly configured Conda environment. After that, select this kernel in VS Code. Make sure your Fabric and Jupyter extensions in VS Code are updated and check the Jupyter output logs for errors like connection timeout. If you still encounter issues, try rolling back to VS Code version 1.98.x, which has shown more stable Fabric integration in user reports.

If this helped solve the issue, please consider marking it 'Accept as Solution' so others with similar queries may find it more easily. If not, please share the details, always happy to help.
Thank you.

[tdevinda]

Thanks for the solution.

I tried to switch the online kernel to 1.2 and use VS code but it did not work.

As for the 1.3 version with the 1.98.x VS code, I did not try. Not in a position to downgrade just for this. Will wait for some stable solution to come out. Until then I think the web version will have to do. 

Thanks. 

[v-hashadapu]

Hi @tdevinda , Thank you for reaching out to the Microsoft Community Forum.

Sorry for the inconvenience. Using the Fabric web UI is indeed a solid workaround now since switching to runtime 1.2 and downgrading VS Code to 1.98.x aren't viable options for you right now.