V1.101.1

Major

• Launched new Budget feature [beta]
• Launched new Audit feature [beta]
• Launched the ability to view unconnected and unmonitored clusters
• Launched the ability to inspect cluster costs on a single pane of glass [beta]
• Future deprecation of Assets API announced, will be moved to Cloud Costs API

Minor

• Improved overview performance for users managing large data sets and multiple clusters
• Added nested navigation
• Added Email alerts for Kubecost & Cluster Health alerts
• Advanced Reports: added a label to cloud breakdown aggregation type
• Network tile on Inspect now sorts traffic in descending order
• Adds a new link to the Namespace Network tile on Overview that shows full network breakdown by namespace
• Improved error handling on Orphaned Resources savings page
• Added CSV as a format option for the /allocation/external API
• Added support for cost difference using new Predict API
• Introduced the /prediction/speccost API for estimating the cost impact of applying a YAML spec to your cluster ahead of time
• Switched 'kubectl cost predict' to use /prediction/speccost 143
• Improved Request Sizing performance by adding server-side paging to Request Sizing results
• Removed unused query param from request sizing V2
• Allow custom minimum CPU and memory params for rightsizing v2 1974
• Continuous request right-sizing will no longer modify Deployments and DaemonSets who currently have > 0 unavailable replicas
• Added kubecostProductConfigs.regionOverrides and corresponding environment variables to chart and deployment template, respectively 2031
• Added OIDC RBAC configs 1979
• RBAC Authorization for OIDC 1222
• Ability to query failed ETL metric Doc

Bugs

• Fixed issue 2003 Custom saved reports via Helm are not functional
• Fixed AWS Discounts (Enterprise + Private Rate Card) missing on /Cloud and /Assets
• Fixed filters not being removed when you use bread crumbs to navigate backwards on allocation
• Fixed an issue in the /savings/requestSizingV2 API in which the endpoint would fail completely with an error like cost per resource hour cannot be <= 0 if the cost for CPU or RAM for any workload is 0 but all other data is valid
• Fixed diagnostics>cloud integration Shows Green check, but detail is failed connection
• Fixed overview timeseries ranges
• Fixed duplicate bug in Advanced Report save
• Fixed bug causing estimated savings on overview and savings page not to match
• Fixed bug that broke URL paths break when using SAML authentication
• Updated Request Sizing icon and wording to increase clarity
• Fixed an issue where filters on the Cluster Sizing page were not showing up for users of Federated ETL
• Fixed an issue where request-sizing page RAM/CPU requested field values change on page refresh
• Fixed multiple Cloud Cost bugs
• Fixed broken link to the diagnostics page
• Fixed an issue where default Share Idle setting was not respected by unconfigured Allocation reports
• Fixed Settings page having backwards Slack and GH links
• Updated dead links on Reports page
• In new prediction endpoints, fixed interpretation of default RAM resource pricing
• Fixed resource cost prediction using only the GPU requested for a single replica. It now uses the GPU requested totaled across all replicas kubecost/kubectl-cost#142
• Fixed an issue causing Diagnostic Notifications (bell in the top right) to show the same message many times
• Updated curl in affected container images to handle CVE-2023-23914

Opencost

Community Contributions

• Added a /pricingSourceSummary API endpoint 1714
• Fixed Panic on Startup with no config file set 1710
• Fixed missing CPU price 1742
• Allow provider regions to be manually configured 1728
• Removing account from key matching between kubecost and cloudcost 1721
• Provides helper functions to access warnings on AssetSets from the parent AssetSetRange 1717
• Fixed ListDirectories for Azure Storage 1707
• Specify units of some CustomPricing fields 1698
• Update CircleCI to work with Docker buildx build 1697
• Fix incorrect value in node vGPU count assignment for vGPUs 1696
• Interfaces needed in Kubecost Asset and CloudCost item for mapping monitored to billed resources 1695
• Ensure /allocation response start and end are always in UTC format 1694

V1.101.0-RC.3

v1.101.0-rc.3 is a pre-release. It is to help gather feedback from the community and give users a chance to test Kubecost before v1.101.0 is officially released.

After taking v1.101.0-rc.3 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 v1.101 release. The estimated release date is Friday, March 10th.

We decided to delay the v1.101 release until Friday, March 10th, to provide more internal testing time and ensure our users a stable release.

V1.101.0-rc.2

v1.101.0-rc.2 is a pre-release. It is to help gather feedback from the community and give users a chance to test Kubecost before v1.101.0 is officially released.

After taking v1.101.0-rc.2 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 v1.101 release. The estimated release date is Wednesday, March 8th.

V1.100.2

Bug fixes

• Fix an unsafe pointer access that, in some cases, was causing panics.
• Reduce the number of large queries on the Overview page, which were causing memory spikes and OOMs for users with large quantities of cloud data.

V1.100.1

Patch notes

• Update an expired SSL cert
• Bump networkCosts version

v1.100.0

Major

• New Kubecost Experience has been promoted to the default view!
• Added Microsoft Teams Alerts
• Ability to predicting cost before deployment + Admissions controller

Minor

• Overview response time improvements

• Overview accuracy improvements

• Allocation Trends API introduced, adds Spend Change percentage to Allocations page

• Added custom labelsupport for Advanced Reports

• UI/UX Update the asset controls to match allocation

• Download cluster sizing recommendation reports as PDF

• Allow users to export saved reports/alerts created via the UI

• Default Allocations Page graph to 'Cost Over Time'

• Improved stability for cluster sizing recommendations

• Enabled annotating the Prometheus server ServiceAccount when using AWS IRSA with Amazon Managed Prometheus

• Cluster node sizing recomendation now take region into consideration

• Launched new more preferment Network page Doc

• Added redirect link after SAML login

• Update chart to use Helm release namespace

  Request sizing Improvements

• Request sizing now supports filters in new UI

• Request Sizing V2 now supports large scale clusters.

• Request sizing columns sortable

• Split the RAM/CPU Requests/Recommended into 4 total columns

• Added support for configurable recommendation query window in continuous request right-sizing via the requests.autoscaling.kubecost.com/recommendationQueryWindow annotation.

• Updates the requestsizing UI to support paging

• Reduced minimum recommended CPU to 10m in the /savings/requestSizingV2 API (and frontend)

• Fixed an issue where the request-sizing page could crash if an AllocationSet was empty

• Fixed multiple drilldown bugs

• Fix CPU cores being misinterpreted as millicores on the top-level request sizing page graph

• Added "Continuous Request Right-Sizing" support for DaemonSets and CronJobs.

  Cloud Cost View Improvements

• Azure CloudCost Integration

• Alibaba Cloud Cost Integration

• Update API types and table values to correctly process API results

• Update Cloud Costs aggregation options

• Update parameters accepted by cloud cost view aggregation.

• Added Cloud Cost support for Advanced Reports

Bugs

• Prevent continuous request right-sizing from infinitely erroring
• Fixed an issue where empty responses weren't handled properly causing Overview page to not load
• Fixed back button logic on drilldown
• Fixed broken links to grafana
• Fixed sorting by efficiency on the allocations
• Enabled copy and paste on drill-in elements
• Fixed an issue where Federated ETL data would not show up on the primary cluster
• Fixed Context Switcher loading and deletion bug
• Fixed inconsistent number of cloud providers
• Fixed details page crash when filtering by label on Allocation
• Fixed Allocation page window results look off near start of cluster's data
• Resolved wild card filtering issues when drilling down
• Fixed Installation of Kubecost on Kubernetes 1.25 fails when installing without Helm
• Fixed Container Request Right Sizing API (v2) with Argo Rollouts CRD fails for 'algorithmCPU=quantile' parameter
• Fixed disabling values reporting breaks bug reporting
• Fixed timeout when trying to capture a bug report

Opencost

• Set up Karpenter capacity type label
• Add support for prefix parameter in bucket config
• Add a unit test to v2 filter parsing for a fully-qualified DaemonSet container query
• Apply node labels to allocation data
• Fix GPU counting bug when separate GPU cost is defined but not CPU/RAM
• Use localhost resolver to get host
• Allocation struct to have Network breakdown values of Zone, Region and Internet costs
• Add endpoints to set and get log levels
• feat(metrics): Add opencost_build_info metric
• perf(query): Align start/end times with step
  

V1.100-RC.1

v1.00.0-rc.1 is a pre-release. It is to help gather feedback from the community and give users a chance to test Kubecost before v1.100 is officially released.

After taking v1.100 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 v1.100 release. The estimated release date is Tuesday, February 7th.

v1.100.0-RC.0

v1.00.0-rc.0 is a pre-release. It is to help gather feedback from the community and give users a chance to test Kubecost before v1.100 is officially released.

After taking v1.100 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 v1.100 release. The estimated release date is Tuesday, February 7th.

v1.99.0

v1.99.0 Release Notes

Major

• Advanced Reports page Beta Launch
• Dynamically applied request recommendations
• Support for Alibaba's public pricing
• Cloud View Alpha
• Recurring Allocation and Assets PDF emailed reports
• Added Orphaned Resources For Azure Users

Minor

• Added network costs and drill-in to the overview page
• Allocation and Allocation PDF Exports
• Customizable email alert subject lines
• 1742 Configure Prometheus scrape metrics
• 1799 RBAC support for authorization to view data by labels
• 1722 Support AWS pricing data in air-gapped environments
• 1477 Add unmounted PVCs to unmounted allocation by namespace
• 1480 Support for on-prem GPU pricing
• Introduce historical efficiency data to the V2 request sizing API
• Changes Inspect header styling to match that of other pages throughout the app
• Support one click request sizing in new UI
• Added an explanatory tooltip to the Name column in the Cost Allocation table to improve user experience
• Notifies users that they are using an unsupported version
• Improved Context Switcher usability
• Added automaxprocs for better performance when CPU requests are set
• Added default request and limits to network costs daemonset to reduce CPU usage under load
• Added a column to the new UI request right-sizing page to show the historical efficiency of the workload
• Improved request sizing drilldown experience
• 1821 Adds Access-Control-Allow-Methods header for the Cluster Controller routes
• Removed the count queries and count features of the new overview page to increase performance
• Improved stability for savings/clusterSizingETL for multi-cluster customers
• Added banner to the new Settings page when an update is available
• Add window to Allocation CSV download
• Reduced log level on spammy disk logs.
• Upgrades Grafana sidecar image version from 1.19.2 -> 1.21.0
• Improve ContainerStats logs for default visibility
• Improve ContainerStats pipeline stability at high Prometheus/Thanos scales by building data in subwindows
• ContainerStats pipeline builds most recent windows first
• Disables ContainerStats pipeline by default (when CONTAINER_STATS_ENABLED is unset) for users who do not use the Helm chart

Bug Fixes

• 1729 Saved reports created via values.yaml ignored if not persistent volume
• Fix a bug where the Cloud Integration details on the Diagnostics page always showed coverage as undefined
• Updated frontend to use new API request for Cost Allocation View API when Idle is set to "Hide" option. (The legacy UI manually filtered out idle data, so the old logic is misleading)
• Fixed bug causing savings total to not match total on overview
• Fix a bug where the UI allowed setting arbitrary windows for alerts, which is not supported
• Fixed shared cost settings to not reflect in Allocation page until hard refresh
• 1642 Beta Solution for failed cluster sizing recommendations when DaemonSets do not run on all nodes
• 1809 Fixed broken URI for keyinstructions section
• 1773 Fixed Installing Kubecost on K8s v1.25 clusters fails due to PodSecurityPolicy
• 1430 Partial fix for Azure cloud integration config ignores kubecostProductConfigs
• 1813 Masks Slack URL in Monitor Cluster Health
• 1339 Fixed Azure nodes naming mismatch
• 1835 Fixed AWS Resource tags in s3 CUR integration have s."resourceTags/user:Name prefix
• Fixed a bug where first load of the application in a new browser could fail to render
• Improved AlertTest error messages

Opencost

• 1541 Fixed ListDirectories for GCS not working properly
• 1539 Fixed a Panic during system disk retrieval for a node when Alibaba service key is not given
• 1537 Fixed UI mismatch between old UI and new UI resulting from the impact of calculating RAM and CPU efficiency of allocations not active for entire duration of the requested window.
• 1531 Updated- golang AWS SDK to v1.44.153 from v1.28.9 as detailed here
• 1530 Reduce log level of metric parsing warnings in ClusterDisks
• 1505 Allow users to define what pricing endpoint to read AWS pricing data from
• 1492 Replace ioutil which has been deprecated since go1.16 and some OS functions are more efficient
• 1480 Adds Custom Pricing for GPU support
• 1477 Added unmounted pvcs to unmounted alloc by namespace
• 1476 Fixed unmounted PVCs being marked as mounted

v1.99.0-rc.9

v1.99.0-rc.9 is a pre-release.

It is to help gather feedback from the community and give users a chance to test Kubecost before v1.99 is officially released.

After taking v1.99 for a test drive, if you have any feedback, it would be great if you could open an issue here.

The official change log will come out with the v1.99 release. The estimated release date for v1.99 is Tuesday, December 13.