Frontend
- Fix an issue where the graph on-hover tooltips in the Cloud Costs page were showing windows in local time, instead of UTC. This sometimes caused the tooltip and x-axis to disagree on date.
- Fix an issue where the Automatic Request Sizing UI in Savings Actions was passing an invalid filter when attempting to filter workloads by Deployment.
- Added a tooltip to the Clusters List page explaining Workload Efficiency, and changed the column name from “Efficiency” to “Workload Efficiency”.
- Fix an issue where the Automatic Request Sizing UI in Savings Actions would sometimes produce malformed timestamps in requests to schedule recurring sizing actions.
- Fix a UI issue where any errors that occurred during the creation of a cluster turndown schedule were not relayed to the user.
- Fix a UI issue where creating an Alert with a window other than 1d-7d would cause it to show up in the Alerts table as having a window of “7d”.
- Note that the UI does not have a way to accurately represent completely arbitrary windows at this time. Setting windows of e.g. “1h” via the Helm chart will produce similarly broken displays in the UI, although the alert will function correctly.
Backend
- Fix an issue where Allocation reports with a step size of “hour” could not be saved.
- Add a DB info log, printed after DB instance creation.
- Improve stability and memory consumption on high scale environments by switching off Jemalloc.
- Fix an issue where idle compute resources would sometimes not account properly in reconciled allocation data.
- Bump image to resolve CVE-2024-35255
- Helm chart
- Add namespace label to aggregator serviceMonitor.
- Add option for aggregator env var for DB_TRIM_MEMORY_ON_CLOSE.
Cluster turndown
- Send HTTP responses in header rather than body for better compliance with HTTP standards.
Cluster controller
- Update Cluster turndown reference version to v0.16.3