Releases: kubecost/cost-analyzer-helm-chart
V1.104.1-rc.0
Minor Updates
- Security Updates
- Update query service replica configuration for large scale deployments.
- Add new fields for namespace labels and annotations in Opencost.
Bugs
- Fix an issue causing Prometheus diagnostics errors.
- Fix an issue with Savings archive storage keys.
- Fix an issue where the Cost Allocation and Cloud Cost charts would show "Other: $NaN" if there were too few items in the chart bar.
- Fix an issue where Cost Center Reports were not applying Cloud Cost filters correctly, and filtering by label was not an option.
- Fix an issue with allocation drilldown when cluster is part of multiple aggregations set.
V1.104.0
Important Notes
- CloudCosts are now enabled by default. GCP users should note that CloudCosts do not support non-Detailed Billing Exports. More information on Detailed Billing Exports can be found here
- Fixed an issue where valid credentials to AWS cloud integrations or Spot configurations would fail
- 2228 Changes all podSecurityPolicy to be disabled by default to prevent Helm issues that will occur if Kubernetes is upgraded to v1.25 while PSP are enabled
Note: Prior to upgrading Kubernetes clusters to v1.25, you must disable PSP used by Kubecost by either upgrading to this release, or for older releases setting the appropriate helm values documented here. Failure to do so will result in Helm being unable to manage the deployment requiring manual intervention.
Major
- Continuous Container & Node Sizing added to automatically ensure your clusters are not overprovisioned
- Namespace turndown added to automatically turn down namespaces when they aren’t needed for use (i.e. nights/weekends)
- New Advanced Report type: Cost Center Reports can create cost centers, or groups of Kubernetes Allocations and External Cloud Costs
- Inspect Shared Costs: Ability to breakdown the costs in the ‘shared’ column of the allocations page
- Auto-Send Reports on an Interval: Have any saved report sent to your inbox with any cadence you like
- Take Cluster Profile into Account When Estimating Savings: More conservative savings optimizations for HA or Prod clusters than Dev clusters
- Great Cloud Integration Experience: Improved UI/UX, ability to delete integrations via the UI, and automatic key verification
- Amortized Costs: Additional cost metric available in the Cloud Cost Explorer
- Overview Load Times: Improved performance of the Overview Page
- Top Traffic Destinations by IP: Visualize the destinations which are being sent the most traffic
- Archivable Savings Opportunities: Hide savings optimizations that aren’t relevant to your organization
- Open Source Cloud Costs: Providing Kubecost’s non-k8s cost monitoring to the Open Source community via Opencost
Minor
- 2245 Explicitly cast query service replicas to int
- 2254 Fix values.yaml for OIDC RBAC
- 1933 Broader support for on-prem PV pricing, including fixing pricing from custom CSVs
Bugs
- Made the Cloud Integrations card fit more nicely with design system
- Fixed Possible Savings on Overview page not matching Savings page Summary
- Fixed regression issue for Alerts
- Repaired weekly ETLs
- Fixed reconciliation not triggering after Asset/Reconciliation rebuilds
- Updated cluster inspect tooltip copy
- Fixed K8s utilization breaking for cloud cost drilldown
- Fixed an issue with sorting request sizing results
- Fixed broken Cloud Cost Filters for Account and Billing Entity
- Fixed Allocation Recurring Update alert types aggregated by cluster or namespace
- Fixed Account-level labels not working
- Fixed a bug where applying filter before selecting aggregation would cause filter to disappear (Cloud Costs)
- Fixed a bug where $NaN would show on the breakdown, filling the remainder cost with an $NaN value
- Fixed a bug where windows larger than the data retention window would break the Cluster Details page
- Fixed high CVE (PRISMA-2022-0227)
- Fixed a bug where custom GCE nodes would break Cluster Sizing Recommendations
- Fixed Request Sizing continually requesting schedules when Cluster Controller was enabled
- 1966 Fixed error for AWS IRSA + Spot Data Feed Users
- 2265 Fixed query-service deployment namespace
- 2278 Removed extra spaces that would breaks the YAML
Community
- 1430 Added current cluster ID filter
- 1832 Added initial ADOPTERS.MD of organizations using OpenCost
- 1869 Refactored Azure provider into pkg/cloud/azure
- 1878 Fixed panic in CSV exporter
- 1897 Fixed panic when metrics are disabled using the disabledMetrics configuration
- 1902 Refactored AWS and GCP providers into their own packages
- 1911 Added governance and maintainers documentation to OpenCost
- 1919 Added
arch
label to cost metrics - 1940 Added support for multiple currencies in the OpenCost UI
- 2206 Added option to configure
securityContext
block on prometheus-server deployment's containers with default of{}
- 2216 Added support for ServiceMonitor relabelings and metricRelabelings
- 2227 Supported priority class in various cost-analyzer components and thanos
- 2241 Fixed secret objects from CSI secret provider
V1.103.5
- Use cached data to improve summary allocation queries for large installations.
V1.103.4
- promoted 1.103.4-rc.0 to a full release
V1.103.4-rc.0
- Resolves issues with specific Helm versions by updating some parts of the Helm chart to make
.Values.kubecostDeployment.queryServiceReplicas
use integers.
V1.103.3
- Cloud Costs
- Fix calculations of item-level modal showing the item's sampled cost over time.
- Update
Billing Account
->Invoice Entity
andWorkspace
->Account
in accordance with the FOCUS spec. - Resolve an issue with filtering by Invoice Entity and Account that resulted from the FOCUS term mismatch.
- Cluster Inspect
- On the cluster inspect list, disable query window options <48h as this list relies on reconciled billing data.
- Fix a bug where individual cluster pages would not load for query windows <48h.
- Fargate
- Fix Fargate node pricing and Fargate pod allocation distributions.
V1.103.2
V1.103.1
V1.103.0
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
- Net Amortized Costs now supported on new Cloud Costs API
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
V1.103.0-RC.9
V1.103.0-rc.9 Pre-release
Edit: V1.103.0-rc.9
Delete: V1.103.0-rc.9
1.103.0-rc.9 is a pre-release. It is to help gather feedback from the community and give users a chance to test Kubecost before v1.103.0 is officially released.
After taking v1.103.0-rc.9 for a test drive, if you have any feedback, it would be great if you could open an issue here https://github.com/kubecost/cost-analyzer-helm-chart/issues/new/choose.
The official change log will come out with the v1.103 release. The estimated release date is Wednesday, May 3rd.