3 May 2021 Release Notes¶
Version 1.1.214
Feature Highlights¶
DataKitchen Alerts Can Support Customer Email Delivery Services¶
The Automation platform can be configured to send alerts when order runs start, produce test result warnings, exceed time limits, fail, and complete successfully. The platform typically sends these notifications using Amazon Simple Email Service (SES) to the email addresses and Slack or Microsoft Teams channels that users set for each kitchen and/or recipe.
With this release, the platform can be set up to send emails through the SMTP protocol to connect with a customer's own email delivery service! Customers who want to pursue this configuration must request the setup through a DataKitchen representative. See Alerts for more information. [DEV-7674, DEV-8201, DEV-8204]
Warning
Changes to Existing Files: The platform will remove the entry "email-delivery-service": "aws-email" from any recipe-level description.json file when you update that file or make changes to recipe alerts via the webapp.
Other Fixes & Enhancements¶
Default Set for Schedule Start Deadline: When users set schedules for their orders, they can set a time period during which the system will attempt to execute the order if it missed its scheduled time. This Start Deadline now defaults to 10 seconds. The system will use the 10-second deadline when the setting is blank or set to any number less than 10; it will apply any values greater than 10. See Schedule Settings for more information. [DEV-7928]
Orders Page Sorting Corrected: Users found that the orders list was not sorting orders chronologically as expected. The problem occurred because the system was not updating the internal timestamp for previous runs of scheduled orders. This issue has been resolved, and order records display from most recent run to oldest, based on the most recent order run of each order. Note that the use of filters may hide some of the order runs for an order, making the sort appear incorrect. [DEV-7785]
Duplicate Runs Prevented: The system was automatically rerunning order runs that failed on out-of-memory errors. These order runs failed properly for requiring more memory than the maximum RAM allocated but should not have been duplicated. This issue has been resolved, and the platform does not rerun these failed order runs. [DEV-8166]
Messaging Improved for Insufficient RAM: A user reported that his order run ended in a hung state when a container node required more memory allocation than the Kubernetes runtime resources had available. Investigation revealed that the system would only fail gracefully in low disk space scenarios. This issue has been resolved, and insufficient memory issues during container node processing will generate the following error in the webapp: "Container terminated. RAM usage exceeded the limit." [DEV-8174]
API Exposes Innermost Exceptions: A user of the Automation API had an issue with a Python script failing, which turned out to be an authentication error. The API was returning a top-level, generic error that was not helpful in troubleshooting. This release updates the API to be able to unpack and concatenate all errors to reveal the inner exceptions for more effective troubleshooting. [DEV-6254]