I have the same problem with Redshift.
Last Wednesday (May 5, 2021) the cluster upgraded from revision 1.0.25813 to 1.0.26073 and the problem started to happen. Here, it affects only a table of a specific external schema that we have. It's a nightmare because we have some queues at RabbitMQ that queries some Redshift data and the consumer dies because the query mysteriously hangs until the it's aborted.
We are now discussing whether to revert the update or wait the solution from AWS.
Hope the AWS devs find the solution quickly.
Edited by: kawe on May 11, 2021 5:00 AM
Hi frecong and kawe,
Have you looked at your queues in Redshift to see if you have a queue wait time issue where the queries you're SQL clients are waiting for might just be waiting on an queue slot to execute. The Redshift doc. can show you how to do this.
Since revision 1.0.26742 the problem seems to be solved. I cannot reproduce hanging queries with or without using cursor. But i cannot find any changelog of that revision. magic things happens.
It's worth noting that Redshift does not communicate a complete change log for Redshift releases, like you may be accustomed to with internal code bases or open source projects. What we get in the release announcements here in the forum or the doc page that points to the forum are the the release highlights that Redshift determines affect or improve the experience of most customers. So, issue resolutions, however severe the issue may be, that only affect a narrow group of customers are often not openly reported or announced. Typically, if you're an enterprise support customer you'll learn of those through your TAM or account team. If you are not an enterprise support customer yet still have support case access and report the issue in a support case you may learn of a fix through post to the support case by the support agent if the support case is still active because the fix is close enough in time to your last interaction with support. Otherwise, for the most part, you, the customer, are left to your own devices to figure out if and when your issue is addressed.
If it is important to you to know when an issue is fixed, it's helpful to have a reproducible test case that you can run periodically or on demand to identify when an issue is resolved. As another Redshift customer, I too find this frustrating at times to be left in the dark on an issue status even with enterprise support and crack account team that stays on top of most every issue. IMO and in my experience, the only way this will change is much like anything else in AWS, if enough Redshift customers voice a strong opinion in favor of change then the AWS "customer obsession" inherent in the AWS value system will kick in and drive a customer driven change. So, please make your voices heard.
Is it possible to have a Redshift Spectrum schema access multiple accounts data ?asked 6 months ago
Is WorkDocs a good fit for transactional/customer-based organization of documentsasked 8 months ago
Can you use a Redshift Spectrum Table in a CTE?asked 8 months ago
Redshift Spectrum: no way to glance at tables with nested dataasked 3 years ago
Prevent Rewrite of Queries in Redshiftasked 4 months ago
Is there a solution for multi-user Notebook on SageMaker?Accepted AnswerEXPERTasked 2 years ago
use Redshift Spectrum to query both Redshift table and s3 fileAccepted Answerasked 2 years ago
Athena and Redshift Spectrum performance best practicesAccepted Answerasked 4 years ago
No Solution Stack named X (via CloudFormation)asked a year ago
Redshift Spectrum with hanging select queries about 25 minutesasked a year ago