Forum Discussion
Fabric Failed to create session for executing notebook
Hi UdaySutar ,
We’re following up on your query to check if the issue has been resolved. Could you please confirm the status?
If the issue has been resolved, we kindly request you to share your resolution or key insights in this thread. This will help other community members facing similar challenges.
If the issue persists, we recommend raising a support ticket instead of connecting via a call. You can submit a request here: Create a Support Ticket.
If we do not hear back, we’ll proceed with closing this thread. However, should you need further assistance in the future, feel free to reach out via the Microsoft Fabric Community Forum and create a new thread. Our team will be happy to assist you.
— Yugandhar
Community Support Team
Hi UdaySutar ,
Step 1: Goto your Workspace settings -> Spark settings -> High Concurrency then enable for notebooks and pipeline run options.
Step 2 : In your notebook attach to high concurrency session(use when you want to run manually)
Step 3: In your data pipeline -> notebook activity -> settings -> Advanced settings -> give your session tag name(anything)
So now when you run your pipeline, if no high concurrency session available it will create high concurrency session and in one session it can execute 5 notebooks.
Regards,
Srisakthi
Hi UdaySutar ,
The error message:
Failed to create session for executing notebook… Cluster was in terminal state=[Cancelled]"This issue often occurs during scheduled executions because the Spark cluster is provisioned on demand. If the cluster fails to initialize in time, it may timeout or terminate unexpectedly.
To improve the reliability of scheduled pipeline runs, consider configuring the environment to use High Concurrency Mode with a session tag. This allows the pipeline to reuse an existing Spark session, reducing cold start delays that often disrupt scheduled tasks.
By using an active high concurrency session, the scheduled pipeline avoids the overhead of starting a new cluster each time, ensuring smoother execution and better stability.
I hope this helps....Please Accept as solution if this meets your needs and a Kudos would be appreciated.
Yugandhar.
