Query runs forever on Synapse Serverless SQL

Edwin 61 Reputation points
2021-11-30T12:48:37.287+00:00

Hi,

we have a problem where OPENROWSET based queries on Synapse Serverless SQL that are now running forever. No errors. One of the queries has now been running for 24+ hours. Any queries that only use sys-tables still work. The issue is the same across all the databases in the same Synapse environment and across multiple different data lakes / storage accounts. West Europe region.

I tried to kill the queries using this:
https://learn.microsoft.com/en-us/answers/questions/77884/how-to-cancel-a-synapse-on-demand-query.html (the one suggested by StefanAzaric-MSFT)
but now the killed queries seem to be in a permanent "KILLED/ROLLBACK" state and new queries still do not work.

Since there is no server to restart etc. with on-demand/serverless sql we are stuck. Can anyone help?

Azure Synapse Analytics
Azure Synapse Analytics
An Azure analytics service that brings together data integration, enterprise data warehousing, and big data analytics. Previously known as Azure SQL Data Warehouse.
4,932 questions
{count} votes

2 answers

Sort by: Most helpful
  1. Edwin 61 Reputation points
    2021-12-01T07:55:56.597+00:00

    We ultimately deleted the Synapse Workspace and recreated it which fixed the issue. I wish there was someway to hard reset / restart the on-demand "Builtin" sql pool for these types of situations.

    3 people found this answer helpful.

  2. Robert Kostecki 0 Reputation points
    2023-05-31T09:10:26.1166667+00:00

    I have faced the same issue with Azure Synapse Serverless when querying data lake based external tables (similar if not the same in nature to the OPENROWSET). The queries were hanging for days. Medium priority ticket was resolved by Microsoft engineers on the next day (pretty fast response).

    The RCA below - it doesn't look like something that could be addressed without help from the engineering team.

    "We identified a FE failover and fixed the issue from our end. Hence, there was a disconnect between the execution nodes which resulted in not generating a query plan and fetching the data remotely.

    This is an intermittent issue and may not occur repeatedly.

    Next Steps:

    We are continuously taking steps to improve the Microsoft Azure Platform and our processes to help ensure such incidents do not occur in the future."

    Many thanks,
    Rob

    0 comments No comments

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.