github kubecost/cost-analyzer-helm-chart v2.7.0
on GitHub

one day ago

Overview

2.7 is considered a production release with a focus on:

  • Diagnostics: ability to quickly see the health of all agents in multi-cluster environments
  • Optimization: See efficiency reports by label and save custom performance profiles to request sizing
  • Collections: Share collections across categories

Known Issues

  • On net new installations, diagnostics UI will appear broken for 15 minutes.
  • The Diagnostics UI is new and may display false positive issues. Use the jump menu to navigate and drill into each cluster to better understand the impact of a given issue.
  • Grafana and Prometheus have unfixed CVEs. These will be patched when upstream projects release GA fixes.

Features

  • Beta: Allow continuous container rightsizing from agent (secondary) clusters with service accounts by @cliffcolvin in #3784
  • Collections sharing: Collections can now be shared across a category.
  • Multi-cluster diagnostics: improved visibility into the health of Kubecost installations on secondary clusters from the primary UI.
  • Change behavior of cost-model to panic when certain misconfigurations are detected. Panic log will point users to fix specific issue. Plus other log fixes.
  • Collect all node labels by default by @avrodrigues5 in #3892
  • Efficiency reports can now be aggregated by workload label in the "Idle By Type" view

Fixes

  • Various image bumps to resolve known CVEs, see current image list below
  • Fix: An issue with ETL Health Checks in diagnostics.
  • Fix: An issue where discounts were not properly getting calculated.
  • Fix: Add /diagnostic/nodeCount to aggregator.
  • Fix: An issue with Idle Load Balancers.
  • Fix: An issue where agent appears healthy when node_total_hourly_cost is not setup in Prometheus scrape.
  • Fix: An issue with uppercase labels in autocomplete queries for assets.
  • Fix: Add first resolved date to multi-cluster diagnostics.
  • Fix: Warning response when failing clusters in cluster sizing.
  • Fix: Add master API key for auth integrations.
  • Fix: Network Insights Ingestion Panics.
  • Fix: Cluster controller helm template by @mittal-ishaan in #3835 and #3886
  • Fix: grafana dashboard templating by @takirala in #3826
  • Prevent installing Grafana dashboards when Grafana is disabled by @chipzoller in #3911
  • Fix forecasting helm to use consistent image/tags by @ivankube in #3863
  • Streamline EKS container image pointers in https://github.com/kubecost/cost-analyzer-helm-chart/blob/v2.7.0/cost-analyzer/values-eks-cost-monitoring.yaml
  • Hide bell by default by @jessegoodier in #3974
  • Fix a cluster controller issue with timeouts on service startup trying to hit Kubecost api.
  • Make continuous request sizing work on secondary cluster hitting Kubecost API url with a service token.
  • Clean up many noisy logs in cluster controller.
  • Fix a bug where help text for request sizing quantile algorithm showed percentages as a fraction of 1% instead of a fraction of 100%.
  • Update network card on allocation details page to indicate that the card is primary-cluster-only. Previously confusing because the rest of the page is mutlicluster.
  • Fix an issue where filtering to a specific cluster on the Savings page was not possible.
  • Fix an issue where in some cases, the Bell icon in the upper-right corner of pages was shown, even in environments that had disabled it.
  • Fix and issue where the Reserved Instances page would crash due to an API change.
  • All documentation links updated to reference IBM documentation site.
  • Fix an issue where clicking through to a cluster from the Clusters page to Allocations would populate a deprecated parameter in the link.
    *Fix a bug where toggling to a different cluster from the Cluster Detail page would not update the Budgets widget.

Other changes

  • Several images are now based on UBI-9 minimal.
  • Add text to the Settings page discouraging use of Share Idle By Cluster in Collections. The setting is deprecated and will be removed in a future release.
  • Remove the "New" tag from External costs and Efficiency Reports navigation items.

Helm Changes

Full Changelog: v2.6.5...v2.7.0

Opencost

  • [#3061] Remove node-label allowlist, so that all node labels can be tied to workloads.
  • [#3080] Add Network Insights to Opencost.
  • [#3085] Fix duplicate labels error in emitted metrics.
  • [#3086] Add metric collectors for promless storage.

Image List

  • gcr.io/kubecost1/cost-model:prod-2.7.0
  • gcr.io/kubecost1/frontend:prod-2.7.0
  • gcr.io/kubecost1/cluster-controller:v0.16.16
  • gcr.io/kubecost1/kubecost-modeling:v0.1.24
  • gcr.io/kubecost1/kubecost-network-costs:v0.17.9
  • ghcr.io/kiwigrid/k8s-sidecar:1.30.2
  • grafana/grafana:11.5.2
  • prom/node-exporter:v1.9.0
  • quay.io/prometheus-operator/prometheus-config-reloader:v0.81.0
  • quay.io/prometheus/alertmanager:v0.28.1
  • quay.io/prometheus/prometheus:v3.2.1
Check out latest releases or
releases around kubecost/cost-analyzer-helm-chart v2.7.0

Don't miss a new cost-analyzer-helm-chart release

NewReleases is sending notifications on new releases.

Get notifications