V1.106.0

Important Notes

• V1.106 is a ‘production’ release focused on targeting bug fixes and stability.
  • More details about our new release structure can be found here.
• The production release includes a significant focus on bug fixes, as well as an extended period of testing and QA process.

Bugs

• Fix NaN warnings in pod logs
• Expose invalid nodes if /clusterSizingETL query fails
• Disable savings cache for Node Turndown
• Add a handler for new node pricing diagnostic API
• Update security dependencies
• Add IP to places where load balancers are created
• Sanitize NaNs for Assets and Allocations
• Fix number mismatch between Allocation UI and Allocation PDF created in recurring update alerts
• Fix recurring Alert set for resource other than primary cluster resource failing to notify user
• Remove ETL Audit Pipeline and API
• Fix multiple “aggregate by” validity of cloud cost report and also shows why the report is invalid
• Add empty Allocation set checks to aggregate store driver
• Remove unused config entries from service
• Fix cluster inspection
• Prevent request sizing crashing
• Move Allocation/Asset/CloudCost/query calls to Query Service
• Cloud Savings: Use APIRootClient for individual Savings calls
• Add ‘Not Contains Suffix’ option to filter builder UI
• Remove unnecessary checks
• Cloud integrations: Adjust fallbacks in form
• Cloud integrations: Show 'delete' icon when editing only
• Add missing translations
• Use budget date range for PDF download)
• Guard against undefined item.type
• Fix Assets NaN
• Use the getCurrentContainerAddressModel() function from utils service in the cluster inspect client (prevents an issue with context addresses getting an extra /model appended to the end)
• Fix overview cloud account links
• Fix Assets range chart; guard against missing data; tests
• Omit inspect menu item for non-inspectable rows
• Fix cluster sizing API query parameter
• Add test for graph-gen's getDataForGraph
• Enable deleting cloud integrations
• Add cloud cost integration error reason in 'Failed' integration status; drops colon from headers
• Add step for configuring IAM permissions
• Fix for doc link scanning and Vercel links
• Move access_key_id from config to separate field
• Drop fetcher_config; makes workgroup optional
• Fix range chart top 10 sorting
• Fix render loop in report save/unsave dialogs
• Add AWS Cloud Integration to Kubecost Cloud
• Drop valueless filters when generating v2 filter string for inspect
• Add Azure logo into Kubecost Cloud UI
• Consider precision for currency format threshold
• Remove 1d-6d menu options
• Update Spot Commander docs links
• Fix Savings pages formatting
• Display the SAML group that the user is in
• Ensure the controller is mapped to controllerName when translating to v2 filters
• Format sub-penny currency values as <0.01
• Update changelogs for 1.106 and bump version
• Simplify how Kubecost determines if a report is selected
• Fix label issue for CloudCosts
• Use the cloud link for cloud-costs for the Report page
• Fix an issue in which the request sizing drill down model sometimes requires two row clicks to open
• Fix title generation issues in Advanced Reporting
• Disable submitting cloud integrations while one is in flight
• Make the proper enablement checks for the cloud costs card, shows the data and links them to the correct page
• Filter conversion tweaks; adds filter and model tests
• Clean up Assets styles to match the other pages
• Add Cloud Cost Reports to Kubecost Cloud
• Fix Cloud Cost Explorer title rendering
• Disable drill down/inspect for unallocated keys/unallocated workloads
• Align the Kubecost Cloud control alignment with on-prem
• Change Cloud Cost Explorer column from ‘K8 Utilization’ to ‘K8s Utilization’
• Sort Cluster UI
• Remove on-prem info and break the title on word so it fits in the modal
• Fix clickthroughs on Overview
• Fix to Cloud Cost Explorer
• Add some messaging around why the integration fields are disabled for GCP integrations
• Fix to Cloud Cost aggregate
• Add missing locale strings
• Fix Allocation query filter
• Fix to Cloud Cost
• Add Cloud Cost Explorer to Kubecost Cloud
• Fix Cloud Cost bug
• Filter empty objects from Allocation data
• Fix a bug where when aggregated by item, accountID filters weren't included in the payload
• Update some placeholder copy
• Fix cloud integration link 404
• Fix initial filter labels issue Assets
• Change currency copy in Settings
• Test SystemNotifications
• Update Actions
• Fix to Cloud Integration
• Fix a bug in which selecting "Last week" or "Last month" resulted in an empty cluster-inspect page.
• Update Alerts query window text
• Delete Budget modal
• Add example text to GCP integration
• Update Cloud config API to match the new accessKey property
• Fix a bug where Window persists after saving
• Update the Reports page
• Add budget form validation
• Increase integration column size
• Swap currency component
• Add federator pod logs to bug report
• Fix Settings page regression
• Fix a bug where Kubecost refetches preview data on each reportName keystroke
• Expose more details for the Azure integration
• Fix a font weight on one of the Overview cards
• Update Savings names
• Fix Allocation drill-in for label
• Use the rawAllocationOnly Max values for the request sizing drilldown table
• Fix resource paths when refreshing on nested routes
• Stop sending sentry errors on 404 XHR requests via Axios
• Cleanup to Cloud Cost page
• Fix incorrect data in request right-sizing CSV vs UI
• Add cloud integrations for Kubecost Cloud
• Show the product UI version in Settings/Diagnostics, not the internal version
• Add query selector for Alert windows
• Bump word-wrap from 1.2.3 to 1.2.4
• Add step to link validation that inverts return code
• Fix disabled check
• Add Allocation filters v2
• Improve suspense handlers and adds an official error page to UI
• Display error on bad JSON
• Add a check that disables the algorithm changer and adds an alert message if the user does not have container stats enabled or the value is undefined.
• Fix to the docs-link validator
• Add a cron workflow for ensuring that no docs links are dead
• Move context switcher to settings
• Fix profile casing on cluster sizing page
• Add dismiss all button to notifications popover
• Pass 'elevated' prop to useButton
• Classic UI switch
• Add @src alias to ease imports that are 3+ dirs up
• Use Reports API on monitoring pages

Community

• 2107 Sanitizes label names before metric emission to prevent label name collisions
• 2093 Adds the creation of lookup key with architecture
• 2083 Fixes parsing of GCP pricing and create new node diagnostic pricing API
• 2076 Fixes subquery fallback to lookup container_cpu_usage_seconds_total metric on container instead of container_name
• 2075 Adds IP field to load balancers
• 2074 Updates Cloud Connections with GetStatus function
• 2070 Adds the Close Stale Issues GitHub Action
• 2069 Stops NaNs from propagating through OpenCost and Kubecost
• 2059 Adds PVs for ProportionalAssetResourceCosts
• 2058 Removes ETL Audit types
• 2056 Fixes PV usage sanity limit
• 2054 Downgrades CSV Export error message
• 2053 Corrects hyperlinks in OpenCost docs
• 2051 Adds Sanity limits for persistent volumes
• 2050 Changes default allocation/compute aggregation to match allocation end
• 2048 Uses the existing variable in error message
• 2047 Fixes the unmounted allocation caused by the offset duration leading to an unmounted PVC coefficient entry
• 2044 Corrects hyperlinks in install docs
• 2039 Reformats OpenCost docs minorly
• 1928 Adds Mimir support with org ID header
• 2522 Fixes issue template about length
• 2521 Fixes issue template file extension
• 2518 Removes Helm 2 manifest
• 2517 Fixes comment misspellings in values files
• 2516 Fixes template
• 2511 Adds imagePullSecret and pod scheduling values to Federator
• 2504 Bumps kiwigrid/k8s-sidecar from v1.23.1 to v1.25.0
• 2498 Updates Grafana
• 2495 Allows Mimir proxy to add basic auth header
• 2494 Allows adding default datasource
• [2493](https://github.com/kub...

V1.105.2

Important note:

• Version 1.105.2 fixes an accuracy issue where Kubecost continued reporting data on nodes even after they were terminated, which resulted in inflated cost metrics. All users running Kubecost 1.105.0 and 1.105.1 are advised to upgrade as soon as possible. This issue impacts both the primary instance and agents.

Upgrade instructions:

helm upgrade kubecost -n kubecost --repo https://kubecost.github.io/cost-analyzer/ cost-analyzer --version 1.105.2

V1.105.1

• Fix an issue in container right-sizing pod details.
• Fix an issue with Azure network costs missing in assets view.

V1.105.0

Important Notes

• Version 1.105.x addresses an issue with Federated ETL in v1.104.x. All Federated ETL deployments should upgrade to 1.105.x as soon as possible. See this doc if you have incorrect data during the time when 1.104 was deployed.
• Environments using Federated ETL: A new deployment will be created that is responsible for aggregating multi-cluster data. This enables better scalability and memory management.

Major

• Fix an issue with Federated ETL where conflicting changes occurred when reconciling cloud costs. Reconciliation is now delayed 36 hours.
• 1985 Added Alias support to v2.1 Allocation filters
• Add persistence to query replicas, allows for dedicated query replicas secret, increased timeout.
• Made /allocation/group a POST endpoint again.
• Enhance the Slack and Microsoft Teams message using go-pretty table.
• Add a new alert type called Cloud Cost Budget to set alerting threshold to various supported CostMetric in CloudCost.
• Switch Asset filters to v2.1 AST.
• Update calls/vars for v2.1 filter Alias support.
• Replace v2 Allocation filters with v2.1 AST
• Sanitize old nil sample AWS access key value.
• Add PUT for an individual group report.

Minor

• Update Storybook to the latest version to resolve a number of CVEs.
• Remove react-code-blocks from codebase.
• Leverage useQueries to make API calls and caching more consistent and cleaner.
• Update the Abandoned Workloads page to use the new UI.
• Add improvements to Continuous Cluster Sizing through the Actions page.
• Add improvements to Guided Sizing.
• Swap insight filter from button with menu to select.
• Update Spot Commander page to new UI style.
• 2419 Fix Network Costs DaemonSet resource limits
• 2399 Add comment to defaultModelPricing
• 2394 Expose port 9004 if OIDC enabled
• 2379 Updat Helm chart to support disabling savings cache
• 2372 Add ability to configure Prometheus retention based on size
• 2370 Add selectors and resources to query service replicas (QSR)
• 2367 Fix (query-service): truncate service name
• 2360 Update QSR config to fix federated data queries.
• 2357 Add option to use federated store over ETL backup
• 2347 Fix missing environment variable for QSR
• 2344 Add ability to optionally use federated store secret
• 2342 Add Advance Network Costs to v0.16.7
• 2331 Add option to disable trends
• 2326 Add extra manifests.
• 2323 Add capability to use AzureAD for OIDC RBAC.
• 2315 Remove federator from cost-model, now runs in own pod.
• 2311 Adjust the template for when users create a bug report in the cost-analyzer-helm-chart repo.
• 2303 Update custom Prometheus docs link.
• 2269 Add an option to add containersecuritycontext and healthcheck probes.
• 2201 Add mimir-proxy to support Grafana Mimir as time series database.
• Update allocation report structure to accept/handle StepSize.
• Reverte Savings marshaling update to prevent data envelope breaking front end.
• Create and used safe-marshaling savings.SummaryResponse type
• Update LBs to work with new OC types.
• Populate templates for email, Microsoft Teams, and Slack.
• Add capability to use AzureAD for OIDC RBAC.
• Add auto-detection for label names in Cloud Cost query filters
• Update QueryServiceOptions to store SharedNamespaces and SharedLabels
• Improve memory efficiency for federator
• Adjust logging for FedETL's Federator and Primary
• Add cluster filter to clustermanagement query

Bugs

• Fix Cost Center Report filters.
• Fix an issue where chart tooltips would show an "Other" category with a cost of $NaN.
• Fix drill-down into labels and filtering by labels on the cloud costs page.
• Fix date display with th instead of rd for double digits ending with 3.
• Fix a bug where the wrong values for cost metrics were sent in cost center reports.
• Handle the clusterSizingEtl call failing more gracefully.
• Fix an issue with Audit sorting.
• Fix cloudCost to AssetSet for VMSS network and Disk resources to be used in reconciliation.
• Fix node type sorting indeterminate results.
• Fix 500 ISEs on external and trends due to NaN/Inf values.
• Fix missing pricing for D*ds v5 instance types.
• Fix aggstore CanQuery fail caused by v2.1 filter.
• Fix group report CPU cost total calculation.
• 1969 Fix an issue when inspecting shared costs with sharedNamespaces and/or sharedLabels set, where the shared costs would not be grouped properly.
• 1966 Fix error for AWS IRSA and Spot Data Feed Users
• [2425] (#2425) Fix Cloud Cost reports in values.yaml not generating a report.
• 1981 Fix Prometheus diagnostics errors
• 2034 Fix the mismatch in sharing namespaces due to unmounted PV cost in each allocation.
• 2031 Fix type field for proportional asset resource.
• 2025 Fix edge cases with AWS CN pricing file inconsistencies.
• 2010 Fix looking up asset field type
• 2003 Fix panic in Coverage IsEmpty/Clone functions.
• 2001 Fix asset query duration
• 1993 Fix nil panics in Azure disk pricing
• 1992 Fix bug in Allocation metrics

Community

• 2035 Add nil checks for allocation field map
• 2023 Put the network entry from Azure CSV into the cloud cost
• 2014 Add default GPU cost
• 2011 Update bingen version tags
• 2006 Ported S3Select Integration to OpenCost)
• 2004 Add mechanism for aggregating metadata across accumulations
• 1995 Switch Assets to v2.1 filters
• 1994 Proportional load bal costs (#2)
• 1982 Switch Allocation filters to v2.1
• 1976 Sanitize old nil sample AWS access key value
• 1973 Added "Container" to Breakdown options for OpenCost UI
• 1972 Close S3 file
• 1971 Implement v2.1 filtering logic for Allocation
• 1970 Add filter v2.1 scaffolding
• 1947 Enable UI image to run as non-root
• 1914 Add securityContext to the OpenCost backend container
• 1896 Introduced new fields, namespaceLabels and namespaceAnnotations, that store namespace labels and annotations, respectively

V1.104.5

• Fixed an issue with Federated ETL where conflicting changes occurred when reconciling cloud costs. Reconciliation is now delayed 36 hours.

V1.104.4

• Add multiple currency support to Opencost.

V1.104.3

• Fix an issue with right-sizing card savings accuracy.

V1.104.2

• Reduce memory consumption in secondaries for large scale clients.

V1.104.1

Minor Updates

• Security Updates
• Update query service replica configuration for large scale deployments.
• Add new fields for namespace labels and annotations in Opencost.
• Add support for federation single db
• Add new helm configuration for query service replicas to use federated data buckets.

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.

V1.104.1-rc.2

• Fix an issue with helm chart values not finding weekly stores in federated data bucket.