Major
- Unreconciled costs can now be highlighted via toggle switch in Settings page
- Ability to multiple query service replicas via Helm config to increase query performance
- Spot Savings Calculation for all eligible workloads
- Ability to filter global savings recommendations and most subpages by cluster
- Updated Budget dashboard to include alerts and improved customization
- Cloud Cost improvements (disabled by default)
- Export Cloud Cost to CSV
- Ability to drill down in Costs
- Added filtering and aggregation to Cloud Costs
Minor
6 months
and12 months
windows are now preset options on Allocations, Assets, and Cloud Cost Explorer dashboards- Added weekly step size to Allocations dashboard
- Added Enable alpha features toggle switch to the Settings page
- Added custom sorting to request right-sizing API
- Update the
/orphaned-resources
route accessible via the Savings page to use the new UI elements and styles - Added Helm logic for providing info for decrypting SAML claims encrypted with Okta. 2118
- Added filtering to the Audits page
- Low disk utilization endpoint
- Added Reconciliation support for Fargate kubernetes usage
- Negative values in the
/request-sizing
savings page should now display as "No savings detected" instead of showing a negative value - Added the ability to download the abandoned workloads savings data as a CSV
- Always set
GOOGLE_APPLICATION_CREDENTIALS
in CC - Added Sorting to Overview page tables
- Users will be alerted when Cluster Controller is not enabled on Actions page. Learn how to enable Cluster Controller here.
- Re-generated ETL files and updated pipeline to only load files that exist, rather than attempt to generate them in the load() function
- Added beta support for Google Cloud Managed Service for Prometheus (GMP) as a prometheus compatible database
Bugs
- Fixed a bug with the
/predict/speccost
API which could cause an unexpected error: "Failed to predict cost impact of specs: cannot predict impact of nil spec" - Fixed UI issue adding AWS cloud integration
- Added empty check for cluster sizing data array
- Fixed error around cloudCost filter key mapping
- Fixed inability to save an advance report
- Fixed a bug where request right-sizing recommendations would fail with a 500 error: “Recommending requests from data: CPU algorithm set to max but no max data is available" when no data matched the user-provided filter. The API now returns a 200 with an empty response instead of an error in this case.
- Fixed a bug where the Overview page was not using the designated Cluster Profile from the API.
- Fixed the broken
/etl/allocation/delete
endpoint - Fixed a bug in Cluster Sizing recommendations in which DaemonSet per-node resource requirements were being overestimated (overestimation would scale with # of replicas).
- Fixed a bug where clicking table rows on the
/low-disk
page wasn't opening the relevant Grafana dashboard. - Updated pagination logic such that only the data for the given page is displayed, instead of all data on every page.
- Added reconciliation for Weekly ETL to fix discrepancy between daily and weekly ETL
Community
- 2079 Added NodeExporter dnsPolicy option