Observe and act

Monitor infrastructure

Kubernetes Monitoring

Release notes

Grafana Cloud

Release notes for Grafana Kubernetes Monitoring

Grafana Kubernetes Monitoring includes these elements, which all have their own releases:

Frontend GUI of the app (also called plugin)
Backend mixin (also called integration), covering the alert and recording rules
Helm chart for the default and other configurations of Kubernetes Monitoring

Frontend release notes

The following are release notes for the Kubernetes Monitoring UI.

3.0.0

10/12/26

Type	Description
Fix	Configuration: Test connection now waits up to 10 minutes for data to arrive, instead of 5. A Cluster that takes a while to start no longer reports a timeout while it’s onboarding normally.
Fix	Nodes: The Node detail page now checks whether a Node is present across the selected time range, and especially at the end of it, so the Node not present message appears consistently.
Fix	Configuration: The Kubernetes Monitoring Helm chart link on the Cluster setup page no longer points to a page that doesn’t exist.
Fix	Security: Updated dependencies (including `dompurify` 3.4.16 and `js-yaml` 5.4.1) to address CVEs.

2.41.0

09/28/26

Type	Description
Enhancement	Jobs: The All jobs page shows a warning when the selected Clusters don’t report the `kube_job_info` or `kube_cronjob_info` metrics, which the Jobs and Cronjobs lists need.
Fix	Workloads: The CPU, Memory, Network, and other detail tabs now load for workloads with very large numbers of Pods (1,000 or more). The queries previously listed every Pod by name and grew too long to run. The Filter Pod selector still lets you choose specific Pods.

2.40.0

09/21/26

Type	Description
Enhancement	Filters: The K8s labels filter is now a menu that groups labels by category, instead of a flat list.
Fix	Tables: Fixed inconsistent cell colors in lists. A column transformation ran out of order and colored the wrong cells.

2.39.0

09/15/26

Type	Description
Enhancement	Tables: On large stacks, lists stay tables and the column picker is always available.
Fix	Configuration: Switching on Pod logs under Advanced logs and events tuning now collects Pod logs. The setting previously wasn’t sent to Fleet Management, so switching it on had no effect.
Fix	Filters: Opening a link that includes a K8s labels filter no longer shows No data. The filter now waits for the Cluster list to load before it runs.
Fix	Security: Updated dependencies (including `grpc` 1.83.2, `qs` 6.16.0, and `fast-uri` 3.1.6) to address CVEs.

2.38.0

08/25/26

Type	Description
Enhancement	Configuration: The deployment code the configuration wizard generates now includes everything Instrumentation Hub needs to instrument services on your Cluster. You no longer need to change your Helm values before using one-click instrumentation.
Fix	Configuration: Test connection no longer fails on a new stack while data is flowing normally.
Fix	Clusters: Selecting a 7-day or 30-day time range on the Clusters list no longer shows an empty page, and the Node count no longer counts Nodes twice. The namespaces list is also faster over long time ranges.
Fix	Nodes: The Nodes list now loads faster and more reliably on large stacks.
Fix	Nodes: The Node labels panel on the Node detail page now shows every label and wraps them to fit the page. Labels previously ran off the right edge of the panel, so values were cut off and some labels weren’t visible.

2.37.0

08/21/26

Type	Description
Enhancement	Links: In trace and log views, Kubernetes attributes such as a namespace or Pod name link to that object’s page in Kubernetes Monitoring. Container names now link too, and links from logs work again.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel search support for workloads.
Enhancement	OpenTelemetry: In preparation for future support of OTel, the home page now recognizes data arriving from OTel collectors, so it no longer reports that no data is flowing.
Enhancement	OpenTelemetry: In preparation for future support of OTel, workload type filters now match workloads reported through OTel, which spell the type differently from Prometheus.
Fix	Jobs: The Logs tab on a Job detail page now shows the Job’s logs. The Pod filter on that tab was empty, so no logs appeared.
Fix	Lists: The workload, namespace, and Node lists now update when you change the time range. They previously kept showing data for the earlier range.
Fix	Panels: The Distribution, Alignment, and Efficiency panels on the CPU and Memory tabs no longer show duplicate Pods and workloads that multiplied with each refresh.
Fix	Configuration: Switching on Cost metrics in the configuration wizard now installs OpenCost, so cost data appears. Switching it on previously did nothing. If you provide your token with Use a stored Kubernetes Secret, cost metrics still aren’t installed.
Fix	Configuration: Custom labels on the Deployment step of the configuration wizard is renamed Local attributes. The attributes you add now reach Fleet Management, where you can search for them and use them to target pipelines. They were previously discarded.
Fix	Configuration: Selecting Auto-discover services on my cluster in the configuration wizard now sets up service discovery. The wizard previously reported success, but Instrumentation Hub never found any services.
Fix	Configuration: Switching on Pod & Service metrics by annotation, Prometheus Operator objects, or Kubernetes manifests in the configuration wizard now collects that data. Switching them on previously did nothing.
Fix	Recording rules: The Kubernetes Monitoring is not functioning fully banner now appears when the alert and recording rules are uninstalled, and stays hidden when they’re installed. It previously got both cases wrong on some stacks. The banner appears only for Admin users, and its button is renamed from Install now to Go to configuration. If you installed the rules yourself rather than from the Configuration page, the banner appears even though the rules are installed.
Fix	Security: Updated dependencies (including `protobufjs`, `dompurify` 3.4.13, and `react-router` 7.18.2) to address CVEs.

2.36.1

08/04/26

Type	Description
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel search support for containers.
Fix	Configuration: Fixed several problems that stopped Test connection from completing on a newly created stack, including the Install Kubernetes backend step running indefinitely, the collector probe returning no data, and the activation step failing when the wizard runs inside another app.
Fix	Panels: Kubernetes Monitoring no longer queries Prometheus before the data sources resolve on page load, which returned `invalidDatasourceId` errors.
Fix	Security: Updated dependencies to address CVEs.

2.36.0

08/03/26

Type	Description
Feature	Configuration: The configuration wizard is now the unified Helm flow, which walks you through backend and distribution, access token, setup choice, monitoring signals, and deployment. This wizard was developed across the 2.34.0 and 2.35.0 releases and is now available. For the steps, refer to Configuration steps for Kubernetes Monitoring with Helm chart.
Feature	Tables: Added a column picker to the Cluster, namespace, Node, workload, Pod, and container lists, and to the tables on detail pages. Hide the columns you don’t need, and each list remembers your choice. Hidden columns aren’t queried, so a long list loads faster.
Enhancement	Assistant: Grafana Assistant now receives the context of the page you’re viewing, including the Kubernetes object you selected, the filters you applied, and the time range, so you can ask about what’s on the page without describing it first.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel search support for Clusters, namespaces, Nodes, and Pods, so objects reported through OTel appear in search results.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel CPU and memory support to the Cluster, namespace, Node, Pod, container, and workload tables.
Enhancement	OpenTelemetry: In preparation for future support of OTel, separated init container CPU and memory requests from regular container requests.
Fix	Workloads: The container allocation values in the Workload CPU and Workload memory panels now count only running Pods. They previously counted Pods in any state, including Pods that had stopped, so expect these values to change slightly.
Fix	Logs: When you change the time range or switch tabs, the Logs and Events tabs no longer show No data for a few seconds before the results appear.
Fix	Clusters: A newly connected Cluster no longer reports that no data is flowing while its metrics are arriving.
Fix	Panels: The CPU requests set, CPU limits set, Memory requests set, and Memory limits set panels on detail pages show their titles in the panel header again, instead of inside the panel beside the container count.
Fix	Knowledge Graph: Health indicators on the Namespaces list now resolve, instead of rendering as empty circles.
Fix	Logs: The Pod filter on the Logs tab no longer drops Pods that stopped running during the selected time range, so Pods from Jobs and CronJobs stay in the list.
Fix	Search: Namespace search results no longer include entries with a blank Cluster name. Selecting one of those entries opened a link that didn’t work.
Fix	Recording rules: The Kubernetes Monitoring is not functioning fully banner, which prompts you to install alert and recording rules, no longer appears before Kubernetes Monitoring is activated.
Fix	Security: Updated dependencies (including `brace-expansion` 1.1.16 and 5.0.7, and `klauspost/compress` 1.18.7) to address CVEs.

2.35.1

07/21/26

Type	Description
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel support to the container Overview page.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel memory support to the Pod Overview page.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel support to the Pods table on the workload Overview page.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel support for the Pods by owner variable.
Fix	Filters: The Kubernetes labels filter no longer adds its label matcher to panel queries, which made panels show no data.
Fix	Filters: The Kubernetes labels filter now discovers and matches external labels, which are keys without the `label_` prefix.
Fix	Filters: The Kubernetes labels filter no longer lists objects from Clusters that don’t match the filter. Object names such as `kube-system` aren’t unique across Clusters, so filtering by name alone returned same-named objects from other Clusters.
Fix	Filters: Lists now show No data when a Kubernetes labels filter matches nothing, instead of keeping the previous results. Broad filters are also faster, and matching is capped at 200 values for each dimension so queries don’t grow large enough to be rejected.
Fix	Workloads: CronJobs no longer disappear from the workload list.
Fix	Recording rules: Restored the earlier recording rule, so panels show data on Clusters that still run an older version of the backend integration.
Fix	Security: Updated dependencies (including `protobufjs` 8.6.6) to address CVEs.

2.35.0

07/15/26

Type	Description
Enhancement	Filters: Added a K8s labels filter to the Kubernetes Overview home page and to the Cluster, namespace, Node, and workload list pages. Filtering on a label, such as `label_topology_kubernetes_io_region=us-east-1`, narrows the page to the objects that carry that label, so every panel and table shows only those objects. The filter appears only where label keys exist, and by default the Helm chart generates them for Nodes only.
Enhancement	Configuration: The configuration page now shows the frontend and backend versions of the Kubernetes Monitoring app.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel CPU and memory query support to the workload Overview page.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel memory query support to the Namespace Overview page.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel support for the workload container status waiting reason.
Fix	Alerts: Pod and container links on the Alerts page now work when the alert has no associated workload.
Fix	Navigation: Fixed the Kubernetes object links from logs and traces not appearing.

2.34.0

07/08/26

Type	Description
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel support to the Workloads list.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel memory query support to the Node Overview page.
Enhancement	Knowledge Graph: Kubernetes tables now pass the full scope (environment, site, and namespace) to Knowledge Graph, so entity health indicators resolve to the correct entity.
Fix	Recording rules: The recording rules banner now rechecks when you change the data source, so it no longer shows an install status left over from the previously selected data source.
Fix	Filters: Dashboard variables (such as the Node filter) no longer reset to All when the container count changes, and panels no longer appear empty after the view switches.
Fix	Alerts: The Node alerts count now reflects only the selected Node, instead of counting alerts from all Nodes.
Fix	Links: Fixed container links on the Pod detail CPU and Memory tabs that could break when workload information was missing.
Fix	Logs: Fixed an error that could break the page while loading logs.
Fix	Security: Updated dependencies (including `undici` 7.28.0 and `dompurify` 3.4.11) to address CVEs.

2.33.3

06/11/26

This release contains dependency updates only. There are no user-facing changes.

2.33.2

06/10/26

Type	Description
Enhancement	Permissions: Activating Kubernetes Monitoring now requires an administrator-level permission, which matches the permission already required to install the backend integration.
Fix	Knowledge Graph: Health annotations on time series panels now stack in lanes instead of covering the graph. Lane rendering requires Grafana 12.4.0 or later.
Fix	Cost: Savings recommendation tables now show the correct Cluster name.
Fix	Configuration: For OpenShift, the generated Helm chart configuration now matches the `node-exporter` label instead of `prometheus-node-exporter`, so host metrics are collected.

2.33.1

06/04/26

Type	Description
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel CPU and memory support to the Cluster Overview page.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel query support to the Namespace CPU and memory tabs.
Fix	Filters: Default Cluster and namespace filters now apply only on top-level pages, so detail pages no longer inherit filters that hide relevant data.
Fix	Configuration: The wizard’s generated Helm values now enable the eBPF profiling type, so profiling works without manual edits.

2.33.0

06/01/26

Type	Description
Feature	Workloads: You can now view a workload’s Kubernetes manifest, compare it against previous versions in a side-by-side diff, and see configuration changes annotated on detail pages, so you can correlate a config change with a change in behavior.
Feature	Alerts: Added an Alerts tab to the Node detail page, so you can see the alerts affecting a specific node in one place instead of filtering the main Alerts page.
Enhancement	Navigation: Added Kubernetes object navigation links from logs and traces, so you can jump from a log line or trace to the related Kubernetes object.
Enhancement	Filters: Added per-user default Cluster and namespace filters, so each user can set the scope they care about without affecting others.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel support to the Namespace table view.
Enhancement	Configuration: Removed the Integration tab from the configuration page to simplify configuration.
Fix	Time range: Fixed time range URL syncing and a stale time picker after drilldown navigation.
Fix	Profiling: Pyroscope flame graph queries now include Cluster and namespace, so flame graphs show data scoped to the resource you’re viewing.
Fix	Security: Updated dependencies (including `qs` 6.15.2) to address CVEs.

2.32.0

05/12/26

Type	Description
Feature	Home page: Redesigned the Kubernetes Overview home page to group active issues by Availability, Stability, and Infrastructure, and added the Efficiency and Alerts tabs.
Enhancement	Assistant: Added inline assistant functionality (beta).
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel support for Node-level CPU and memory usage and capacity metrics.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added Cluster-level memory support for OTel data sources.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added a podOwner multiexporter to support OTel at the Node level.
Enhancement	Profiling: Flame graphs on workload and Pod detail pages now open to an available profile type (for example, memory) instead of defaulting to CPU and showing “no data” for services that don’t emit CPU profiles. Also reduced redundant Pyroscope queries during dashboard auto-refresh.
Fix	Configuration: Fixed the configuration wizard’s generated Helm command failing on Helm v3. The command now uses `--atomic` instead of `--rollback-on-failure`, which is Helm v4 only.
Fix	Configuration: Fixed the Kubernetes Secret name in the wizard-generated Helm command to match the name expected by the Helm chart, so the chart can find the credentials Secret.
Fix	Security: Updated dependencies (including `uuid` 14.0.0 and `postcss` 8.5.14) to address CVEs.

2.31.1

05/05/26

This release contains internal code cleanup only. There are no user-facing changes.

2.31.0

05/04/26

Type	Description
Enhancement	Configuration: Added ArgoCD to configuration. You can now generate an ArgoCD Application manifest directly from the configuration wizard, alongside the existing Helm and Terraform options.
Enhancement	Search: Added a Containers filter to the Search page, with autocomplete support for container name and image.
Enhancement	Search: You can now refresh search results manually or set an auto-refresh interval.
Enhancement	RCA workbench: Added a Workbench button to pages for adding entities to RCA workbench. Hovering over entity ring icons show a Workbench button.
Enhancement	OpenTelemetry: In preparation for future support of OTel, added OTel metric support for homepage count panels (clusters, namespaces, nodes, pods, workloads), workload/namespace/node/Pod name variables, and node-level CPU and memory requests and limits.
Enhancement	Profiling: The profiling section on Pod and workload detail pages is now collapsible, and the collapsed state persists across pages.
Enhancement	Configuration: Added a banner on every page when alerts and recording rules aren’t installed, with an Install Now link to the configuration page.
Enhancement	Configuration: The configuration wizard now generates values for Helm chart 4.0 or later.
Fix	Configuration: Fixed display issues in the configuration dialog box opened from the integrations page.
Fix	Nodes: Fixed the AWS EC2 button not returning results on the Nodes page.
Fix	Workloads: Pod count panels on the Workload detail page now display whole numbers, making time-series values easier to read.

2.30.1

04/22/26

Type	Description
Enhancement	CPU metrics: Added CPU efficiency stats (requests/capacity, usage/capacity, usage/requests) to the Cluster detail page’s CPU tab. These metrics lay the foundation for OpenTelemetry support.
Enhancement	CPU metrics: Added CPU requests and limits to Pod and container detail pages. These metrics lay the foundation for OpenTelemetry support.
Fix	Cluster CPU tab: Fixed extra table columns incorrectly appearing as links in the Cluster CPU tab.
Fix	Logs: Fixed log and event queries not refreshing when the time range changes.
Fix	Alerts: Fixed the Namespace detail page showing alerts from other clusters with the same namespace name.
Fix	Security: Updated Go to 1.26.2 and updated dependencies to address multiple CVEs.

2.30.0

04/14/26

Type	Description
Feature	Alerts: Added an Alerts tab to the Cluster detail page.
Enhancement	Knowledge Graph: Added Knowledge Graph health annotations to time series panels on detail pages (Cluster, Namespace, Node, Pod, and workload), showing detected issues as highlighted time regions.
Enhancement	Pyroscope integration: Fixed the “View profiles” button and flame graphs. This was due to not finding Pyroscope data when the container name differs from the workload name. When a workload has multiple containers with Pyroscope data, a drop-down menu lets you choose which one to view.
Fix	Init containers: Fixed native sidecar containers (for example, Istio) showing no data in resource views. Kubernetes registers these as init containers, and the UI was only querying regular container metrics.
Fix	Alerts: Fixed Grafana-managed alerts not appearing on the main Alerts page. Only data source alerts were displayed.
Fix	Energy: Fixed the workload Energy tab showing incorrect data because queries weren’t filtering by namespace.
Fix	Predictions: Fixed the “Predict CPU usage” and “Predict memory usage” buttons rarely working. They only succeeded on fresh page loads and failed after navigating between pages.
Fix	Configuration: Fixed the “Create token” button being disabled during onboarding. The button required the backend to be installed first, but the backend requires a token to install, so users couldn’t proceed.
Fix	Security: Updated dependencies to address multiple CVEs.

2.29.0

04/02/26

Type	Description
Feature	Search: Added contextual subtitles (Cluster, Namespace, workload) to search autocomplete results, making it easier to distinguish between resources with similar names.
Fix	CPU metrics: Fixed Node CPU usage showing roughly twice the actual value when both node-exporter and `kubelet` report the same metric, because the query summed both instead of removing duplicates. Also fixed an incorrect CPU usage/capacity calculation on the Home page.
Fix	Energy: Fixed broken links on Energy pages when navigating from Node to Pod or from Cluster to Namespace.
Fix	Configuration: Fixed the configuration page showing disabled buttons for Kubernetes Monitoring admins who weren’t also Grafana org admins. The page now correctly uses the Kubernetes Monitoring plugin role for access.
Fix	Security: Updated dependencies to address multiple CVEs.

2.28.2

03/25/26

Type	Description
Fix	Continuous profiling: Fixed profiling time range to use the dashboard time range instead of a hardcoded 24-hour lookback, and disabled the “View profiles” button when the specific profile type doesn’t exist for a workload.
Fix	Continuous profiling: Fixed Pyroscope service name matching for `namespace/workload` format when multiple service names share the workload name as a substring.
Fix	Configuration: Renamed configuration wizard step labels to match instructions (“Stack information” to “Stack and platform”, “Platform and features” to “Monitoring type”).

2.28.1

03/16/26

Type	Description
Enhancement	Continuous profiling: Added Pyroscope integration with embedded flame graphs on workload and Pod detail pages, and direct links to Profiles Drilldown from CPU and Memory tabs.
Enhancement	Configuration: Renamed and clarified all four configuration wizard step labels for consistency, and reworded the step two instructions to better match the workflow order.
Fix	Search: Fixed search page not loading previous search terms when opened from a shared or bookmarked URL.
Fix	Configuration: Fixed the Integration Hub onboarding flow where pasting an existing access token did not populate the token in the generated `helm upgrade` command.
Fix	Configuration: Fixed the OpenCost documentation link not displaying as a clickable link in the cost metrics alert.
Fix	Security: Updated dependencies to address CVEs.

2.28.0

03/11/26

Type	Description
Feature	Configuration: Redesigned the configuration page into four steps for a clearer setup experience.
Feature	Health: Added a new Health page showing a live snapshot of active issues across all Clusters.
Enhancement	OpenTelemetry: As part of adding support for OpenTelemetry, added CPU metrics for OTel data sources in Node, Namespace, and Cluster views.
Fix	Error messaging: Improved error messages for recording rule operations to show the actual error instead of a generic failure message.
Fix	Metrics accuracy: Panels now filter out internal bookkeeping metrics reported by the container runtime (cAdvisor), so CPU and memory values reflect only your actual application containers.
Fix	Workload metrics: Used workload-level recording rules instead of Pod filtering for more accurate workload metrics.
Fix	Panel queries: Added unique query reference IDs to prevent data conflicts in panels with multiple queries sharing the same runner.

2.27.2

03/04/26

Type	Description
Fix	Predictions: Dynamically load prediction modules to prevent page crashes if a prediction module fails to load.

2.27.1

03/03/26

Type	Description
Feature	Search: Redesigned the search page for better user experience.
Enhancement	OpenTelemetry: As part of adding support for OpenTelemetry, added CPU limit metrics for OTel data sources in Namespace, Node, and Cluster views so you can compare CPU usage against configured limits.
Enhancement	Configuration: Added a disabled state to feature toggles that aren’t supported on the selected platform.
Fix	Logs: Fixed Loki data source picker to only show Grafana Cloud Loki data sources, which previously could show non-Grafana-Cloud data sources that wouldn’t work correctly.
Fix	Events: Added a dedicated query for `object_name` and `object_kind` fields in event logs so that event details display correctly.
Fix	Workload detail: Renamed “Pods” label to “Pods in time range” to more accurately reflect the metric.

2.27.0

02/23/26

Type	Description
Enhancement	OpenTelemetry: Added support for OpenTelemetry `kubeletstatsreceiver` CPU metrics.
Fix	IOPS by Pod: Fixed write IOPS to use correct `container_fs_writes_total` metric.
Fix	Pod filtering: Fixed page breaks with many Pods by using recording rule instead of URL parameters.
Fix	OpenTelemetry events: Fixed events labels not displaying correctly for OpenTelemetry events.
Fix	Configuration: Changed default Cluster name to ‘my-cluster’ in onboarding instructions.

2.26.0

02/10/26

Type	Description
Feature	Manage costs: Redesigned the costs and savings view.
Enhancement	Logs: Use node filter (if available) for OTel logs in node logs view for more accurate filtering.
Enhancement	Configuration instructions: Added cluster name to the simplified Alloy Helm instructions for better clarity.
Fix	Alerts: Removed node filter in pods table alerts to prevent incorrect filtering.

2.25.3

02/03/26

Type	Description
Feature	Search: Added ability to search by container image name, making it easier to find all instances of a specific image.
Enhancement	Alert detection: Enhanced alert detection to support both Prometheus Alertmanager-based alerts and Grafana-managed alerts during the alerting platform migration.
Enhancement	Knowledge Graph integration: Added banner to enable and navigate to Knowledge Graph for root cause analysis when not enabled.

2.25.2

01/27/26

Type	Description
Feature	Pod count: Added Pod count panel to workload detail pages.
Fix	Logs: Fallback to default logs query if no data found in any of the queries.

2.25.1

01/27/26

Type	Description
Fix	Logs: Fallback to OTel logs queries as expected.
Fix	Security: Updated dependencies to satisfy CVEs and updated eslint.

2.25.0

01/20/26

Type	Description
Feature	Logs: Replaced logs with new drilldown logs extension for improved log exploration.
Enhancement	Network metrics: Added network metrics for Windows nodes.
Fix	Logs: Use query runner as data provider to get the logs query.
Fix	Security: Updated @remix-run/router to v1.23.2 to fix vulnerability.
Fix	Pod metrics: Removed by label filter in Pod level request and memory sets for more accurate metrics.
Fix	Cronjob: Fixed cronjob filtering to properly filter by namespace.

2.24.3

12/11/25

Type	Description
Feature	Configuration: Added the capability to enable or disable forwarding traces to application receivers.
Enhancement	Changed the default view in the Logs and Events tab from the table view to the logs list view.
Fix	Jobs: Fixed Reason column in the Pods table under Jobs.
Fix	Jobs: Fixed an issue where empty or missing schedules could cause errors when displaying Cronjob schedules.
Fix	Workloads list: Fixed workload counts on the namespace list so that they now accurately reflect the number of workloads in each namespace.

2.24.2

11/20/25

Type	Description
Feature	EC2 instances: Added links showing which EC2 instances correspond to which Cluster nodes.
Enhancement	Explore link: The Explore link opens in a new browser tab, offering stability of the GUI with large datasets.
Fix	Jobs: Fixed job URL creation to use the job name instead of owner kind, ensuring more accurate links for Kubernetes jobs.

2.24.1

11/20/25

Type	Description
Fix	Knowledge graph: ObjectDetailExtension now uses the Kubernetes `workload` concept instead of `entityName`. This ensures displayed details are more accurate and terminology matches current Kubernetes standards.
Fix	Jobs: - Fixed job URL creation to use the job name instead of owner kind, ensuring more accurate links for Kubernetes jobs.

2.24.0

11/13/25

Type	Description
Feature	Access control: Added the ability to use RBAC roles.
Fix	Jobs: Assigned a default owner if no owner type exists, to ensure job links work properly.

2.23.0

10/30/25

Type	Description
Feature	Grafana Assistant: Added Grafana Assistant icons to most panels and added the Investigate with Assistant button to all detail pages.

2.22.0

10/22/25

Type	Description
Feature	Jobs: Added jobs and cron jobs monitoring.
Enhancement	Filtering: Enabled column filtering on all tables.
Enhancement	Activation: Changed name of activation tab to Manage app.
Fix	Workbench: Removed filter from columns with a checkbox.
Fix	Node table queries: Used node filter for accuracy.
Fix	Column headings: Removed incorrect units.

2.20.0

8/29/25

Type	Description
Enhancement	Integrations: Added a Kubernetes filter to the Connections page.
Enhancement	Configuration: Added the ability to activate Kubernetes Monitoring.

2.19.8

8/21/25

Type	Description
Enhancement	Integrations: Removed integrations from the Configuration page.

2.19.7

8/19/25

Type	Description
Enhancement	Configuration: Removed the selection in configuration to switch on Application Observability now that users can activate it from the product.

8/5/25

Type	Description
Enhancement	Asserts: Added new term ‘insights’ instead of ‘assertions’.
Fix	OTel: Updated the cluster events query to improve support with OpenTelemetry Collectors.
Fix	Scrape intervals: Removed hard-coded time interval so that panels reflect custom scrape interval.

2.19.5

7/25/25

Type	Description
Enhancement	Asserts: Added health indicator icons to lists of components.
Fix	Pod CPU and memory: Reverted fix from 2.19.3 due to bugs it was causing.
Fix	Container memory: Fixed container memory panels to aggregate correctly.
Fix	Cluster list: Corrected the number of nodes for a cluster in the info section.

2.19.3

7/16/25

Type	Description
Enhancement	Cost overview: Added ability to jump to list of Clusters from Cost Overview page.
Enhancement	Integrations: Added a message on the Integrations page that integrations will soon be accessible from the Connections page.
Fix	Pods: Include all Pods if no pod value specified and use podName value if available on Pod CPU and Memory queries to prevent no data from showing.

2.19.2

7/3/25

Type	Description
Feature	GPU: Added GPU tabs and panels to Cluster and Node pages.
Feature	EC2 link: Added link to navigate between EC2 instance in Cloud Provider Observability and Kubernetes Monitoring.
Enhancement	Cost: Added tooltips for cost panels to clarify underlying calculations and their time frame.
Enhancement	User experience: Streamlined descriptions of panels for clarity and brevity.
Fix	Storage: Changed inode values to percentages.
Fix	Workloads: Fixed query to ensure all barepods workloads are listed, whether desired Pods or ready Pods.
Fix	CPU and memory tabs: Corrected the calculation for the Efficiency Usage/Requests panel on the CPU and Memory tabs of the Node detail page.
Fix	Argo Rollout Pods: Removed duplicate query to prevent incorrect count.
Fix	Container page: Fixed tab links to navigate to appropriate pages.

2.19.1

6/11/25

Type	Description
Enhancement	Serverless and Windows Nodes: Added support for showing Node memory usage on platforms that cannot use Node Exporter (such as GKE Autopilot and AWS EKS Fargate), and Windows.
Enhancement	Logs table: Added the ability for a user to rollover a line in the logs table and click an icon to see the full log message.
Fix	Queries: Fixed queries for Cluster names with a `.`, which causes results to be empty.
Fix	Non-Kubernetes Nodes: Prevented the app from selecting Nodes that are not Kubernetes Nodes to be counted and displayed.

2.19.0

6/4/25

Type	Description
Enhancement	Memory: Added a memory tab and panels to Node, Namespace, Workload, Pod, and Cluster detail pages.
Enhancement	Workloads: Added the capability to view nonstandard workloads that contain cronjobs, bare pods, static pods, Argo Rollouts, Strimzi Pod sets, and others.
Enhancement	Logs: Created a table format to display logs and events.

2.18.0

5/21/25

Type	Description
Enhancement	CPU tabs: Added CPU tabs to the Pod, Workload, and Namespace detail pages.
Enhancement	GKE Autopilot and AWS EKS Fargate: Used `kubelet` Resource Metrics (`node_cpu_usage_seconds_total`) to show CPU usage for platforms that cannot use Node Exporter, including GKE Autopilot and AWS EKS Fargate.
Fix	Link to Pod: Fixed ability to go to a specific Pod from the Network Bandwidth by Pod panel on the Node details page on the Network tab.
Fix	URL: Fixed variables not synching with URL.
Fix	Metrics status page: Corrected spacing between panels.
Fix	DNS: Added a `.` to the end of hostnames to prevent DNS confusion.
Fix	OTLP: Included the OTLP Gateway destination in the Kubernetes Helm chart to route metrics, logs, and traces using the OTLP format.
Fix	Outlier detection: Changed Outlier detection to use podNames to prevent incorrect counts of outlier Pods.

2.17.0

5/6/25

Type	Description
Feature	CPU details: Added CPU tab for Node details page and Cluster details page.
Fix	Table headers: Made table headers visible for light mode.
Fix	Pods: Fixed Pods with the same name on different Clusters to be differentiated from each other.
Fix	Table columns: Make it more apparent table columns can be resized.
Fix	Data accuracy: Ensure that CPU and memory requests only look at active pods, not completed or failed.
Fix	Terraform files: Ensure use of strings for usernames in Terraform files to ensure compatibility for Helm chart values.

2.16.0

4/23/25

Type	Description
Enhancement	Storage: Added IOPS and throughput panels on all Storage tabs.
Fix	Cost: Prevented duplicate cAdvisor scrape jobs from affecting cost allocation.
Fix	Alerts: Fixed namespace drop-down menu on Alerts page from not being populated.
Fix	Pod names: Corrected formatting of Pod names when using Explore from Pod lists.
Fix	Memory usage: Prevented completed and pending pods from affecting memory usage queries on the Pod details page.

2.15.1

4/17/25

Type	Description
Enhancement	Alerts: Added capability to filter alerts on Alert page by Node.
Enhancement	Outlier detection: Outlier detection for memory automatically runs when you visit a workload details page for multi-Pod workloads, and links to Explore to show the outlier Pod(s) using more memory.
Fix	Integration: Correct the configuration snippet for the Core DNS integration.

2.15.0

4/14/25

Type	Description
Feature	Outlier detection: Outlier detection automatically runs when you visit a workload details page for multi-Pod workloads, and links to Explore to show the outlier Pod(s) using more CPU.
Enhancement	Detail pages: CPU and memory columns and prediction buttons for each row have been added to the Cluster, workload, and namespace detail pages.
Fix	Events and logs status check: Fixed status check and Explore links for events and logs on the Metrics status tab.
Fix	Overview page: Fixed the All alert links on the home page.
Fix	Overview page: Corrected the workload count on the overview page to more closely correlate to the workload count shown on the workloads list page.
Fix	Fleet Management: Fixed reference to Fleet Management Terraform variables.

2.14.1

4/2/25

Type	Description
Enhancement	Performance: Decreased the training and inference latency of the machine learning functionality to show results quicker.
Enhancement	Alerts: Added filtering by alert name.
Enhancement	Overview page: After filtering by Cluster or namespace, you can click on the count of Clusters, Nodes, namespaces, or workloads and show only those objects within the scope of the filter.
Enhancement	User experience: On Workload and Namespace list pages, made the Cluster names clickable and added graphs to the counts.
Fix	Workload counts: Corrected calculations on workload counts on the Overview page.
Fix	List pages: Restored the alerts table column.
Fix	Alloy logs instance: Set the Pod security context for the Alloy logs instance on OpenShift.
Fix	OpenShift: Corrected the port names for OpenShift so that Alloy can discover targets to scrape.
Fix	EKS Fargate: Removed mounting of /var/log and /var/lib/docker/containers.

2.14.0

3/24/25

Type	Description
Feature	Troubleshooting: Added a debug panel showing the source of a panel’s metrics, available from the panel menu.
Enhancement	Performance: Added a view for each list page when queries extend beyond performance capabilities which degrade page loading. This view includes a message to search or filter for specific components.
Fix	Performance: Stopped unrelated panels from reloading when the logs data source is changed.
Fix	Accuracy: Stopped counting containers that are not running, which caused CPU limits and requests to appear higher than they should.
Fix	Platforms: Disabled unnecessary components for specific platforms: nodelogs on Fargate, and profiling, Beyla, and Windows Exporter on Autopilot.

2.13.2

3/18/25

Type	Description
Fix	Fleet Management: Fixed interaction with Fleet Management so that Alloy installation shows positive on Metrics Status page.
Fix	Fleet Management: Changed configuration to store secret for Fleet Management in secret instead of environmental variable.

2.13.1

3/13/25

Type	Description
Fix	Removed `stabilityLevel: public-preview` in relation to Alloy from the code generated on the Configuration page.
Fix	Prevented a race condition causing HTML error messages to appear.
Fix	Asserts: Restored RCA workbench linking.

2.13.0

3/9/25

Type	Description
Feature	Configuration: Added toggle in easy deploy for users to add Fleet Management for management of Alloy deployments.
Enhancement	Cluster detail page: Added buttons to allow users to navigate to workloads and namespaces within the cluster.
Fix	Memory queries: Switched to Working Set Size (WSS) instead of Resident Set Size (RSS) to increase accuracy of memory statistics and match `kubectl` and Linux Out-of-Memory behavior.

2.12.3

2/20/25

Type	Description
Fix	Configuration and status tabs: Fixed issue of configuration state not persisting when user switches between the tabs.
Fix	URL stability: Maintain URL on refresh of page.
Fix	Integrations: Updated integrations to work with Kubernetes Monitoring Helm chart version 2.

2.12.2

2/6/25

Type	Description
Fix	Workload details page: Fixed TypeError to prevent unsafe access to data source.

2.12.1

2/6/25

Type	Description
Fix	Receiver URLs: Updated URLs to match the version 2 Alloy deployment.
Fix	Tempo and Logs: Updated configuration to fix Tempo protocol and autopilot for logs.

2.12.0

2/5/25

Type	Description
Feature	Storage: Added storage tabs and graphs to Cluster, Node, workload, namespace, and Pod detail pages.
Enhancement	Configuration: Updated the easy deployment configuration page to use Helm chart v2.
Enhancement	Recording rules: Added the install/uninstall button to the Workload Recording Rule status for the cloud-provisioned hosted metrics data source.
Fix	Prediction button: Disabled if the time range selected is less than two hours.
Fix	Data source drop-down: Removed machine learning metrics data source from the drop-down menu.
Fix	Data source drop-down: Added verification that the data source is valid prior to displaying in the drop-down menu.
Fix	Workload list page: Fixed to not display all columns if filters result in more than 1,000 results.

2.11.2

1/15/25

Type	Description
Fix	Time range: Ensured the time range selected in Kubernetes Monitoring is transferred to Application Observability when the View application layer button is clicked.
Fix	Performance loading: Resolved issues related to rerendering of pages.
Fix	Filters: Ensured when time range is changed that all filters are not returned to the default.

2.11.1

1/9/25

Type	Description
Enhancement	User experience: Added the tool tip to the Copy icon.
Fix	Node count: Fixed incorrect Node count on Kubernetes Overview page due to empty node labels.
Fix	Performance loading: Improve loading performance of tables in detail pages.
Fix	Memory data: Rollback to the original recommendation for memory requests to equal memory limits.

2.11.0

1/6/25

Type	Description
Enhancement	Performance: If Workload list page result is over 1,000 workloads, the page will load quickly but no usage or cost data will show. A message will display suggesting the list be filtered, and include a link to search.
Fix	Metrics status: Detect duplicate container metrics.
Fix	Data source: Filter the data source drop-down menu to be limited to Grafana Cloud data sources.

2.10.0

12/9/24

Type	Description
Feature	Performance: Improve performance by performing a code split.
Fix	Loop error: Fixed a history loop error and page breaking.

2.9.0

11/25/24

Type	Description
Feature	Search: Added the ability to search for any Kubernetes object.
Fix	Metrics status: Show the status checks even if no clusters were detected.

2.8.8

11/12/24

Type	Description
Fix	Erroneous no data error: Added a null check to prevent the no data message from displaying due to UID not found for a database.

2.8.7

11/6/24

Type	Description
Fix	Asserts: Fix to ensure Kubernetes Monitoring is aware Asserts is enabled.

2.8.6

11/5/24

Type	Description
Fix	Asserts: Fix the time range so that it is synchronized between Kubernetes Monitoring and Asserts.

2.8.5

10/30/24

Type	Description
Fix	Updated link to the Application Observability GUI.
Fix	Properly align Cluster columns.

2.8.4

10/17/24

Type	Description
Enhancement	User experience: Created Cluster, namespace, and Pod columns for alerts on home page.
Enhancement	User experience: On the Metrics status tab, added the ability to select the status of all or many Clusters.
Fix	Instead of displaying no data on the home page should an outage occur, the home page will show historical data prior to the outage and current data post-outage.
Fix	GUI: Fixed various anomolies with the GUI, including the location of the navigation drop-down menu and the width of certain columns.

2.8.3

9/23/24

Type	Description
Fix	Alerts: Hide panel menu item for creating alerts if a user is not an administrator.

2.8.2

9/12/24

Type	Description
Enhancement	User experience: Restricted the namespace field for the quick configuration process to only valid characters.
Enhancement	User experience: Added the link to the Cluster below the title on detail pages.
Enhancement	Troubleshooting: Added the Helm chart to the Metrics status page.
Enhancement	User experience: Added waiting reasons and colors for Pods that are either waiting due to startup or due to a problem.
Enhancement	Troubleshooting: Added the waiting reason for Pods that are waiting instead of running.
Enhancement	Alerts: Added the ability to create an alert item from a panel.
Fixes	Fixed the container memory limit on Container detail pages to be accurate. Ensured logs and events are not listed when there are no Pods. Fixed ability to remove a ’no data’ filter when a time range is changed.

2.8.1

8/25/24

Type	Description
Feature	New platform: Added Azure AKS as a separate platform for configuration.
Enhancement	Metrics status: Added statuses for recording rules and KSM job label.
Enhancements	User experience: Moved up the required backend installations step to prevent misconfigurations. Restricted integrations view to administrators only and improved message to non-administrative user. Added links to documentation on Overview, Cost, Alerts, and Configuration pages. Adjusted the width of filters and moved copy button closer to item for ease of use. Refresh home page when data starts being sent.
Enhancement	Code snippets for integrations: Added configuration code snippets for Apache CouchDB, Apache Hadoop, Apache Tomcat, and IBM DB2.
Fix	CPU statistics: Fixed issue that was causing the memory legend to show in relation to CPU stats.

2.8.0

8/6/24

Type	Description
Feature	Energy use: Added energy metrics by using third-party Kepler open source software.
Fix	Duplicates: Deduplication of Pods and containers.

2.7.0

7/29/24

Type	Description
Enhancement	User experience: Added a tab for logs and events on every detail page, including the Cluster detail page.
Enhancement	Integrations: Added snippets for the Alloy and Varnish integrations.
Enhancement	Logs: Made log queries work with OTel-style labels.
Fixes	Correct time range format on home page. Ensured Kubernetes Monitoring opens in correct time range from Asserts.

2.6.1

7/16/24

Type	Description
Enhancement	Alerts: Added the ability to click on the alert name from the Kubernetes Overview page.
Enhancement	Integrations: Added configuration snippets for Squid and Apache Mesos.
Fix	Prevented continuous rendering of data in the Safari browser.

2.6.0

7/10/24

Type	Description
Feature	Home page: Upgraded the home page to better show the status of healthy Kubernetes objects as well as any potential issues.
Feature	Asserts button: Added Assertions button to details pages.
Enhancement	User experience enhancements: Added quick ways to navigate within the app, including the drop-down menu for all main pages, the overview and network tabs, deep linking from the home page CPU and memory graphs to the Cluster, improved Helm chart configuration process, and visual tags to indicate the type of Kubernetes object for each view.

2.5.0

6/25/24

Type	Description
Feature	Network panels: Added Network Bandwidth and Network Saturation data.
Feature	Cost overview panels: Added a 90-day view of total compute cost, average cost per Pod, and average Pod count on the Cost Overview page.
Enhancement	OpenTelemetry: Added support for displaying events coming from the OpenTelemetry Kubernetes Events Receiver.
Fix	Workloads with same name: Corrected CPU and memory usage data for workloads with the same name but in different Clusters.

2.4.0

6/18/24

Type	Description
Feature	Deleted objects: Added the ability to find deleted Clusters, Nodes, Pods, containers, workloads, and namespaces.
Enhancement	User experience: Improved “no data” messages on home page.
Enhancement	Metrics: Host metrics generation.
Fix	Error messages: Fixed query to prevent an invalid error message.
Fix	Update host metrics switch language and disable CRDs on OpenShift.

2.3.5

6/12/24

Type	Description
Enhancement	Integration: Added Apache Airflow configuration snippets for integration.
Fix	Duplicates: Removed duplicate data showing on home page.
Fix	Alerts: Improved alert count accuracy for certain kinds of Pod types.

2.3.4

6/7/24

Type	Description
Enhancement	Added time range selector to home snapshot page.

2.3.3

6/3/24

Type	Description
Enhancement	Integrations: Added configuration snippets for the integrations Wildfly and OpenSearch.
Enhancement	Performance: Improved load time for queries spanning longer time ranges.
Fix	Refresh and Cluster filters: Fixed so refresh does not cause filters to be removed.
Fix	Cardinality page: Ensured the Series per namespace column data link is accurate.
Fix	Documentation links: Updated and corrected documentation links on Kubernetes Monitoring main page (that is shown prior to configuring).
Fix	Workload type: Ensured different workloads that are similarly named accurately show the type of workload and the data for the workload.

2.3.2

5/21/24

Type	Description
Enhancement	Performance: Improved loading performance of CPU and memory usage tables.
Fix	Added a null check for provider to avoid errors on Cluster list page.

2.3.1

5/16/24

Type	Description
Enhancement	Integration: Added configuration snippets for the Apache Solr integration.
Fix	Fixed Helm chart integration snippets to be compatible with Alloy.
Fix	Fixed recording rule to include ReplicaSets, static Pods, and bare Pods so that missing workloads now show and incorrectly labeled workloads are now labeled correctly. This also fixes workload usage and cost data.

2.2.5

4/25/24

Type	Description
Enhancement	Static Pod and Pod type: Added the static Pod label to all static Pods and the Pod label to bare/unmanaged Pods .
Enhancement	Integrations: Added flow configuration snippets for Velero and MS SQL Server integrations.
Fix	Fixed link to alert rule after automatic refresh.

2.2.4

4/22/24

Type	Description
Enhancement	Data source persistence in URL: Persisted the Loki data source in the URL so that accurate links can be shared.
Fix	Added check to prevent query errors.

2.2.3

4/18/24

Type	Description
Feature	Refresh control: Added refresh and auto-refresh choices to all pages.

2.2.2

4/17/24

Type	Description
Feature	Alerts: Added the Alerts column to the Nodes and Cluster list pages.
Enhancement	URL data source persistence: Persisted Prometheus data source in the URL so that accurate links can be shared.
Enhancement	Integration: Added configuration snippets for OracleDB integration.

2.2.1

4/15/24

Type	Description
Feature	Alloy update: Replace Grafana Agent with Grafana Alloy.
Feature	Cardinality: Added the Cardinality page to replace the Kubernetes Cardinality dashboard.
Feature	Alerts: Added the Alerts column to these pages: Pod detail, Pod list, and Workloads list.
Enhancement	Integration: Added configuration snippets for TensorFlow Serving integration.
Enhancement	User experience: Added the ability to select multiple clusters for filtering.
Fix	Removed logs panel from Node detail page for future refactoring.
Fix	Corrected improper indent in configuration snippet of IBM MQ integration.
Fix	Added `kube_deployment` as well as `kube_replicaset` to ensure all workload types are collected.

2.2.0

4/8/24

Type	Description
Fix	Refactored workloads to use more efficient queries.

2.1.3

4/4/24

Type	Description
Enhancement	Added the workload type when available to filter the list of Pods to improve search performance.
Fix	Enhanced the multi-line regular expression to support a broader collection of logs.

2.1.2

4/2/24

Type	Description
Fix	Reverted from support for OTel log labels.

2.1.1

4/2/24

Type	Description
Feature	Added interactivity with Sift automated machine learning investigation tool.
Enhancement	Added configuration snippets for the Couchbase and IBM MQ integrations.
Enhancement	Added support for OTel log labels.
Fix	Combined `target_info` with `traces_target_info` metrics to ensure all applications are found.

2.1.0

3/25/24

Type	Description
Enhancement	Configuration: Added configuration snippets for the PgBouncer integration. Added link to `extraConfig` information in the Helm documents from the code snippet areas of an integration.
Enhancement	Dashboards: Removed dashboards due to Angular deprecation.
Enhancement	Messaging: Improve wording of default messages when a data source is updated for a view.
Fix	Persist Cluster URL values to correct broken links.

2.0.4

3/13/24

Type	Description
Feature	Asserts workbench: Added RCA Workbench button to allow navigating to Asserts.
Enhancement	Explore: Expanded the explore feature to the list of all Pods. Improved the user experience for exploring data.
Fix	Removed inaccurate timestamp for namespace creation on Namespace details page.
Fix	Corrected the query for phase state, which is shown by expanding a list item on the Namespaces page.
Fix	Add the `http` prefix to OTLP endpoints on the Cluster configuration tab when OTel receivers are selected.
Fix	Persisted the data source between views.

2.0.3

3/6/24

Type	Description
Feature	Added the capability to navigate to Application Observability.
Feature	On the Cluster list page, you can click the Explore button to view the queries used in each column of the usage and cost data. This enables you not only to understand the data, but debug any issues or create your own queries or dashboards
Feature	Added support for the Istio integration.
Feature	Added a warning icon and tooltip on column headers of lists to indicate error detection and query failure.
Feature	With the easy configuration using the Helm chart, added the capability to enable OTel and Zipkin receivers.
Enhancement	Added the filter toolbar to list tables for better user experience.
Enhancement	Added duplicate metrics detection to Node Exporter, Windows Exporter, OpenCost, cAdvisor, and `kubelet`. When there is an error, a yellow warning icon appears next to the item on the Metrics status page.
Enhancement	Refined queries for “OOMKilled” containers to prevent counting of duplicate containers, specifically when a container is being killed and another is being created in its place.
Enhancement	The link to examine an alert opens in a new tab so that context within Kubernetes Monitoring is not lost.
Enhancement	Updated URL so that any shared URL includes any variables and the time range selected.
Enhancement	Release note text in the app links to the release notes in documentation.
Enhancement	Breadcrumbs now include page tab names.
Enhancement	Corrected the query for unallocated resources (PV) panel.
Fix	Created an additional check for deep links related to breadcrumbs display.
Fix	Refined queries to ensure deduplication of metrics, including removing reliance of `uid` label and having node/instance joins also use the Cluster label.
Fix	Corrected queries for the Explore button on the Workloads page.
Fix	Fixed CPU usage prediction graph to display correctly within window.

2.0.2

2/15/24

Type	Description
Enhancement	Refined deduplication of KSM `kube_node_status_capacity`, and added support for node with or without trailing port numbers.
Fix	Created more accurate cost calculations in the case of mixed `node` and `instance` labels.

2.0.1

2/15/24

Type	Description
Enhancement	Added the workloads column to the list of namespaces.
Fix	Corrected the calculation for CPU average queries.

2.0.0

2/6/24

Type	Description
Feature	The Alerts page provides a centralized location to view all alerts related to your infrastructure and the applications running within it. Graphs show alerts firing by Cluster and namespace, as well as by alert severity. The alert name and severity also display on the home page in the list of Pods in trouble. On the Namespaces and Workloads tabs, a column for each item shows the amount of firing alerts.
Feature	The ClickHouse integration is available for use with Kubernetes Monitoring.
Enhancement	On the home page, you can click the Cluster count to navigate directly to the list of Clusters on the Clusters tab, and click the Node count to navigate directly to the list of Nodes on the Nodes tab.
Enhancement	On usage cost/efficiency tables, the tables are sorted by default alphabetically by the name of the item (Cluster, namespace, and so on).
Enhancement	To improve loading performance, the namespaces, workloads, and Nodes lists are filtered by default by the first Cluster in the filter list.
Enhancement	You can click the Explore button on any item on the Metrics status tab to troubleshoot configuration issues and customize queries for status.
Enhancement	To prevent deployment failure due to timeout, increased the timeout for the Helm chart to five minutes.
Enhancement	An informational banner displays on the Dashboards page to indicate the prebuilt dashboards are being deprecated and removed, with the exception of the Cardinality dashboard.
Fix	Resolved a duplicate series error on the container details page.
Fix	Improved the status panel failure detection for kube-state-metrics.
Fix	When you click Start Sending Data, the Configuration page displays.

1.7.0

1/22/24

Type	Description
Feature	You can view usage and cost graphs and data on the Pod details page.
Feature	When using EKS on Fargate, Pod logs are made available by using the Kubernetes API to gather them.
Feature	Added or updated support for these integrations: Aerospike, ActiveMQ, Cassandra, Apache HTTP Server Project, cert-manager, ClickHouse, CoreDNS, GitLab, Kafka, MongoDB, NGINX, and PrestoDB .
Enhancement	On the Configuration page, you can add an expiration date to the token and view the scope of the token.
Fix	Corrected the following broken links: workload links on the Pod list of the Node details page, Cluster status link on the Configuration page, and links when Cluster name contains a slash character.

1.6.2

1/10/24

Type	Description
Enhancement	User experience enhancements: The home page displays the home page name on the breadcrumbs, and filter options display alphabetically.
Enhancement	You can copy any object in the table lists by clicking the copy button within the row.

1.6.1

12/19/23

Type	Description
Feature	IBM Cloud is available as a platform that you can select when using the configuration wizard.
Enhancement	You can choose Loki from the data source selector in all detailed views.
Enhancement	A link to documentation for understanding CPU limits is available on detail pages showing CPU limits set.
Enhancement	All IEC units show in standard form.
Enhancement	Throughout Kubernetes Monitoring, you have the copy icon available to copy any text so that you can use it in your own query, use it in the command line, send it to others, and so on.
Fix	Fixed empty Cost panels for customers who use forward slashes in Cluster names by changing to pipe formatter.
Fix	Corrected inconsistencies in the number of Nodes displayed by using an instant query for any single value result instead of using a range query. This improves app performance and data accuracy.
Fix	Fixed accuracy of CPU usage data on the Node details page.
Fix	Improved accuracy of machine learning prediction by using query variables to dynamically set parameters for calculation.

1.6.0

12/10/23

Type	Description
Feature	Cluster detail page: Added optimization panel showing CPU and memory utilization, and cost information for the time range selected.
Feature	Node detail page: Added optimization panel showing CPU and memory utilization, and cost information for the time range selected.
Enhancement	Cost page: All panels on the Cost page provide the Explore button so you can view the query, debug issues, or create your own dashboard.
Fix	Storage savings: Fixed storage results, which displayed multiple series errors in cases where a PersistentVolumeClaim was present on more than one node within the selected time range.
Fix	Container details page: Fixed events, which weren’t displaying.

1.5.0

11/29/23

Type	Description
Feature	Container detail page: Added page. This page includes CPU and memory utilization, and cost information. The restart and termination panels are adjacent to each other to allow quicker correlation of restart and termination events.
Feature	Namespace detail page: Added optimization panel showing CPU and memory utilization, and cost information.
Enhancement	Workloads detail page: Added optimization panel showing CPU and memory utilization, and cost information. Updated the feature for detecting outlier CPU usage in the Pods to uses the time range selected in the time selector instead the latest/current time.
Fix	Pod detail page: Updated the CPU and memory queries for accuracy. Removed duplicate containers that were sometimes listing.
Fix	Home page: Decreased load time for the Pods in trouble panel, and filtered only for issues that are currently active.
Fix	Added Zipkin port configuration to ensure traces work properly.
Fix	Updated internal release process and documents.

1.4.1

11/10/23

Type	Description
Feature	Configuration wizard: Added a selector for the platform.
Feature	Pod detail page: Added an optimization panel containing CPU and memory utilization and cost.
Fix	Time range selector: Changed to use the time zone of the user, to ensure accuracy of time range data.
Fix	Use new metrics receiver flow component in the integration code snippets.
Fix	Fixed missing data source for k3d environment.

1.4.0

11/3/23

Type	Description
Enhancement	Time range selector: Added to all appropriate pages to allow a specific time range for data to be displayed.
Enhancement	Metrics status page: Improved for better user experience.
Enhancement	Configuration wizard: Improved visibility of OpenCost configuration information.
Enhancement	Pod details page: Used Scenes to build panels on this page.
Fix	Default instructions in Github for integration: Fixed the appropriate name for Prometheus component.
Fix	Error message for testing integration: Revised for validity and comprehension.
Fix	Workload key and data list fixes: Revised to use correct key on workloads list.
Fix	Kafka integration: Revised job name to kafka for integration to work properly.
Fix	Status queries: Adjusted to show consistent results.

Backend release notes

The following are release notes for the Kubernetes Monitoring backend.

2.3.2

July 2026

Type	Description
Enhancement	Alert rules: Updated kubernetes-mixin to v1.5.6. `KubePodNotReady` now also fires for Pods that are running but report not ready, so it catches readiness probe failures. It no longer fires for Pods that are already ready, or for Pods that are intentionally held back from scheduling with scheduling gates.

2.3.1

May 2026

Type	Description
Fix	Alert rules: Updated kubernetes-mixin to v1.5.4. `KubeHpaMaxedOut` now matches on the `cluster` label, so it evaluates correctly when more than one Cluster sends data to the same stack.

2.3.0

April 2026

Type	Description
Enhancement	Alert rules: Updated kubernetes-mixin to v1.5.0. Adjusted `KubeHpaMaxedOut` to not fire when an HPA’s minReplicas equals maxReplicas, which is common during incident response or load testing. No new alerts are introduced; upstream `*InstanceUnreachable` alerts are excluded from this release because they would cause false positives with the current configuration.

2.2.4

February 2026

Type	Description
Enhancement	Alert detection: Re-enabled support for Grafana-managed alerts. Alert queries now check both Prometheus Alertmanager (`ALERTS` metric) and Grafana-managed alerts (`GRAFANA_ALERTS` metric) to ensure all alerts are detected. During the transition period, you may temporarily see duplicate alerts for the same condition.

2.2.3

January 2026

Type	Description
Enhancement	Updated kubernetes-mixin integration to 1.4.2.

2.2.2

December 2025

Type Description

Enhancement Alert rules: Updated KubePodNotReady to remove Failed from phase label filter; updated KubeContainerWaiting from aggregated metrics using sum to using raw metrics; updated KubeCPUOverCommit and KubeMemoryOvercommit to differentiate between Non-HA and HA clusters; updated KubeQuotaAlmostFull, KubeQuotaFullyUsed, KubeQuotaExceeded to do explicit join instead of ignore; updated CPUThrottlingHigh to deduplicate with topk; updated KubeletPodStartUpLatencyHigh with topk for deduplication.

Enhancement Recording rules: Updated scheduler_e2e_scheduling_duration_seconds bucket to scheduler_scheduling_attempt_duration_seconds_bucket (recording rule name also updated); updated scheduler_binding_duration_seconds_bucket to scheduler_pod_scheduling_sli_duration_seconds_bucket (recording rule name also updated); updated the kubelet PLEG node_quantile:kubelet_pleg_relist_duration_seconds:histogram_quantile recording rule to deduplicate with max by.

2.2.1

December 2025

Type	Description
Fix	Temporarily disabled Grafana-managed alerts.

2.2.0

June 2025

Type Description

Enhancement Alert rules: Added the cluster label to any alert description where it was missing. Added KubePdbNotEnoughHealthyPods, KubeNodePressure, and KubeNodeEviction. Updated KubeletTooManyPods to deduplicate Pod counts and improve evaluation performance. Updated KubeStatefulSetReplicasMismatch to also fire when Pods cannot be created. Updated KubeNodeNotReady and KubeNodeReadinessFlapping to filter out cordoned Nodes. Removed KubeAPIDown and KubeletDown alerts as they are misleading.

Enhancement Recording rules: Added node_namespace_pod_container:container_cpu_usage_seconds_total:sum_rate5m. Updated namespace_workload_pod:kube_pod_owner:relabel to support more workloads. Updated all apiserver rules to support both Prometheus v2 and v3 histogram bucket formats. Updated apiserver_request:availability30d to prevent null values. Deprecated node_namespace_pod_container:container_cpu_usage_seconds_total:sum_irate.

2.1.2

December 2024

Type	Description
Enhancement	Alert rules: Made the `KubeDaemonSetRolloutStuck` alert configurable. Added instance label to `KubeAggregatedAPIErrors`. Added cluster label to `KubeClientCertificateExpiration` description.
Enhancement	Build process/tooling: Introduced automated Go dependency updates. Security updates CVE-2024-45337 and GO-2024-3333.
Fix	Recording rules: Corrected a data calculation in cluster_verb_scope:apiserver_request_sli_duration_seconds_count:increase.*.

2.1.1

November 2024

Type	Description
Enhancement	Alert rules: To better support multi-cluster environments, added `cluster` label to `KubeAPIErrorBudgetBurn`, `KubeAPITerminatedRequests`, and `KubeClientCertificateExpiration`.
Enhancement	Recording rules: Improved data consistency for `cluster_verb_scope_le:apiserver_request_sli_duration_seconds_bucket:increase1h`, `cluster_verb_scope_le:apiserver_request_sli_duration_seconds_bucket:increase30d`, `cluster_verb_scope:apiserver_request_sli_duration_seconds_count:increase1h`, and `cluster_verb_scope:apiserver_request_sli_duration_seconds_count:increase30d`.
Enhancement	Build/process tooling: Updated dependencies to address vulnerabilities.

2.1.0

September 2024

Type	Description
Enhancement	Added alert groups: - `kubernetes-storage` (five `KubePersistent.*` alert rules) - `kube-apiserver-slos` (four `KubeAPIErrorBudgetBurn` alert rules) - `kubernetes-system-apiserver` (six system API Server alert rules, including `KubeAggregatedAPIDown`).
Enhancement	Added recording rule groups: - `kube-apiserver-..rules` (32 `apiserver.` recording rules) - `kubelet.rules` (including `node_quantile:kubelet_pleg_relist_duration_seconds:histogram_quantile`) - `kube-scheduler.rules` (nine `cluster_quantile:scheduler_.:histogram_quantile` recording rules).
Enhancement	Updated upstream Kubernetes mixin: Split `k8s.rules.container_resource` group into four groups to improve performance.

2.0.0

March 2024

Type	Description
Enhancement	Removed dashboards from Kubernetes Monitoring now that all views are available in the application.

1.1.2

January 2023

Type	Description
Fix	Removed duplicate recording rule group.
Fix	Corrected broken links on Home dashboard.

1.1.1

November 2023

Type	Description
Fix	Disabled “all” and multi-select on Events Data Source variable.
Enhancement	Split `k8s.rules` group into seven groups to improve evaluation performance.

1.1.0

October 2023

Type	Description
Feature	You can focus on cardinality exclusively related to Kubernetes by using the prebuilt Kubernetes Monitoring Cardinality dashboard.

1.0.0

October 2023

Type	Description
Enhancement	Added `job` label to `KubePodNotReady` alert to be compatible with new Grafana Alerting structure.

0.0.14

July 2023

Type	Description
Feature	Added recording rules for Windows Exporter, for compatibility with node-exporter queries.
Enhancement	Restored events selector to prevent logs from being returned as events.
Enhancement	Updated upstream Kubernetes mixin.

0.0.13

May 2023

Type	Description
Enhancement	Added ability to use any (non-empty) job label on dashboards panels and rules.
Enhancement	Added Node Exporter to configuration status panel.
Enhancement	Added duplicate installation detection to configuration status panel.
Enhancement	Added informational tooltip to configuration status panel.

0.0.12

April 2023

Type	Description
Fix	Updated upstream agent-operator jsonnet-libs to inherit a fix for relabeling related to hitting a labeling limit with Loki.

0.0.11

March 2023

Type	Description
Fix	Corrected template variables on Efficiency dashboard.

0.0.10

February 2023

Type	Description
Fix	Removed the beta tag from the Events panel on the homepage dashboard.

0.0.9

February 2023

Type	Description
Fix	Corrected Cluster aggregation for some recording rules.

Helm chart release notes

The following are release notes for the Kubernetes Monitoring Helm chart.

For the full list of historical releases, refer to Grafana Kubernetes Monitoring Helm chart releases.

4.5.x

Covers releases 4.5.0 through 4.5.3.

Type	Description
Feature	Added the `prometheusMetricsReceiver` feature, which lets applications push Prometheus-formatted metrics to the in-cluster Alloy for forwarding to any metrics destination.
Feature	Added the `lokiLogsReceiver` feature, which lets applications push Loki-formatted logs to the in-cluster Alloy for forwarding to any logs destination.
Feature	Application Observability: Added OTLP HTTP receiver timeout and keep-alive settings.
Enhancement	Integration instances and profiling sub-features can each be routed to a different collector by setting `collector` on them.
Enhancement	Destinations can overwrite cluster labels. This is enabled by default for Pyroscope and OTLP destinations, which keeps the existing behavior.
Enhancement	Added the native histogram scrape options to all Prometheus Operator object types (PodMonitors, Probes, and ServiceMonitors).
Enhancement	Added `kubelet_evictions` to the default `kubelet` allow list, and `kube_pod_status_ready`, `kube_poddisruptionbudget_status_current_healthy`, and `kube_poddisruptionbudget_status_desired_healthy` to the default kube-state-metrics allow list, so the provisioned alert and recording rules have the metrics they need.
Enhancement	Updated Alloy Operator to 0.8.0, kube-state-metrics to 8.6.0, OpenCost to 2.5.31, and Node Exporter to 4.59.0.
Fix	Fixed control plane (kube-proxy, kube-controller-manager, and kube-scheduler) and etcd integration scraping on IPv6 and dual-stack Clusters.
Fix	Fixed duplicate configuration block declarations when the same data processor is used by more than one feature on a collector.
Fix	Fixed the incorrect `job` label on the Node logs feature.
Fix	The cluster name for Beyla is set only when `cluster.name` is used, instead of being set to an empty string.

4.4.x

Covers releases 4.4.0 and 4.4.1.

Type	Description
Feature	Added the `windowsEventLogs` feature, which gathers Windows event logs from Windows Nodes, along with `windows`, `windows-host-process`, and `windows-scrapeable` collector presets for scheduling Alloy on Windows Nodes.
Feature	Added the `router` destination, which routes telemetry to different destinations based on the value of a resource attribute or label (default `k8s.namespace.name`). Unmatched data goes to the required `defaultDestinations`.
Feature	Host metrics: Windows host metrics can be collected directly by Alloy with `hostMetrics.windowsHosts.source: alloy`, instead of deploying a separate Windows Exporter.
Enhancement	Host metrics: `linuxHosts`, `windowsHosts`, and `energyMetrics` each accept their own `collector`, so Linux and Windows host metrics can be collected on separate DaemonSets.
Enhancement	Regular expressions can be used in the `namespaces` list for many features.
Enhancement	Added `attachNamespaceMetadata` to the annotation-based discovery feature, which makes namespace labels available for relabeling rules.
Enhancement	Added a validation that detects a host port conflict before Beyla deploys. Disable it with `autoInstrumentation.beyla.portConflictCheck: false`.
Enhancement	Cluster metrics: When `kubeScheduler.discoveryMode` is `eks-proxy`, collects the EKS control plane resource metrics `kube_pod_resource_request` and `kube_pod_resource_limit`.
Enhancement	Updated Beyla to 1.16.11 and kube-state-metrics to 8.4.0.
Fix	The Prometheus destination’s cluster label now honors `cluster.nameFrom`.
Fix	Host metrics: Honors `windowsHosts.scheme` and `windowsHosts.bearerTokenFile` when scraping an external Windows Exporter.

4.3.x

Covers releases 4.3.0 through 4.3.3.

Type	Description
Feature	Added the `otel-receiver` collector preset, which opens the standard OTLP receiver ports on the collector (4317 for gRPC and 4318 for HTTP).
Feature	Added the `nop` destination, which drops all telemetry sent to it.
Feature	Pod logs via OpenTelemetry: Added `filters` to drop Pod logs based on the Pod’s annotations or labels.
Feature	Cluster metrics: Added a `discoveryMode` option to `kubeScheduler` and `kubeControllerManager`. Set it to `eks-proxy` to scrape the control plane from the Amazon EKS Control Plane Metrics API.
Feature	Application Observability: Added `overrideUnknownServiceNames`, which overrides the `service.name` attribute when its value is `unknown_service*`.
Feature	Database Observability: Added support for custom labels.
Enhancement	A destination can be disabled with `disabled: true`, which removes it from all destination lists, Secrets, and generated Alloy instances. This makes it possible to remove a destination defined in a shared values file.
Enhancement	Added the opt-in `alignServiceNameWithOTelSemConv` flag to align `service.name` and `service.namespace` on logs and profiles with the OpenTelemetry Kubernetes semantic conventions that Beyla uses, so telemetry correlates across signals.
Enhancement	Added `global.image.registry`, `global.image.pullPolicy`, and `global.image.pullSecrets`, which apply to the Alloy instances, the Alloy Operator, and the Alloy removal hooks.
Enhancement	Added `extraResourceAttributes` and `extraResourceAttributesFrom` options, and resolver and timeout settings for `processors.serviceGraphMetrics`, to the OTLP destination.
Enhancement	Application Observability: Added `connectors.spanMetrics.metricsExpiration` and `connectors.spanMetrics.seriesExpiration`, so stale metrics and series expire instead of accumulating.
Enhancement	Added a validation that each collector’s Alloy stability level is high enough for the experimental or public preview components in its configuration.
Enhancement	Added a validation that detects another Alloy Operator already running in the Cluster. Disable it with `alloy-operator.conflictCheck: false`.
Enhancement	When `alloy-operator.deploy: false`, the Alloy image tag is pinned so an external operator doesn’t substitute its own default image. Set `image.tag` on a collector to override.
Enhancement	`replaceComponent` entries are validated across collectors and can be marked optional.
Enhancement	Added a `presets` attribute to the remote configuration collector attributes.
Enhancement	Updated Alloy Operator to 0.6.3, kube-state-metrics to 8.1.3, Node Exporter to 4.56.1, OpenCost to 2.5.29, Windows Exporter to 0.12.8, and the SDK Injector to 0.2.1.
Fix	Standardized on upper case for log level labels.
Fix	Added missing metric processing rules for the Istio integration.
Fix	Service integrations: Validation now catches an integration with no collector assigned.
Fix	Fixed chart schemas where the only valid type was `null`.

4.2.x

Covers releases 4.2.0 through 4.2.2.

Type	Description
Feature	Introduced data processors for transforming data between features and destinations. Initial processors are `custom`, `kubernetesEnrichment` (copies namespace and Pod labels and annotations onto metrics, logs, traces, and profiles), and `ec2Enrichment` (copies AWS EC2 instance tags).
Feature	Host metrics: Added `hostMetrics.linuxHosts.source: alloy` to collect Linux host metrics directly with Alloy instead of Node Exporter. This requires a privileged DaemonSet with the `linux-host-monitor` and `daemonset` presets.
Feature	Added the Grafana SDK Injector and the Beyla Kubernetes cache as deployable `telemetryServices`.
Enhancement	Deprecated the `metricEnrichment` option on the Prometheus destination in favor of the `kubernetesEnrichment` data processor. The option still works.
Enhancement	Added the `root`, `host-network`, `host-storage`, `host-cgroup`, and `host-tracefs` collector presets, and deprecated the `privileged` preset. `privileged` still works but can’t be combined with the new presets.
Enhancement	Pod logs: Added `attachNamespaceMetadata`, which makes namespace labels available as `__meta_kubernetes_namespace_label_*` for relabeling.
Enhancement	Service integrations: Added per-instance `extraDiscoveryRules` to enrich an integration’s metrics with custom labels before scraping.
Enhancement	Changed the default remote configuration `pollFrequency` from `5m` to `30s`, and added the `CLUSTER_NAME` environment variable to remote-configuration collectors.
Enhancement	`auth.type: basic` is inferred for remote configuration and destinations when a username and password are set but `auth.type` isn’t.
Enhancement	Node logs: journald priority is used as the log level when the log message doesn’t contain one.
Enhancement	Added a `source="spanmetrics"` resource attribute to metrics generated by the span metrics connector.
Enhancement	`telemetryServices.beyla` and `telemetryServices.sdkInjector` now default to `deploy: false` in the top-level values, so they don’t deploy unintentionally on Helm versions affected by a subchart value regression.
Enhancement	Updated Beyla to 1.16.10, Alloy Operator to 0.6.1, kube-state-metrics to 7.8.1, Node Exporter to 4.56.0, and OpenCost to 2.5.26.
Fix	Prometheus destinations keep the `service.name`, `service.namespace`, and `service.instance.id` resource attributes on `target_info` as labels. `openTelemetryConversion.keepIdentifyingResourceAttributes` now defaults to `true`. Set it to `false` to restore the previous behavior.
Fix	OTLP destinations no longer produce `;`-joined duplicate label values on `target_info` when a scrape target exposes its own `target_info` metric, and derive the correct service identity for metrics converted from a Prometheus scrape, which fixes service-to-logs correlation in Application Observability.
Fix	Beyla auto-instrumentation no longer discards the default `discovery` targets when you set another field under `discovery`, and `autoInstrumentation.beyla.service.targetPort` now propagates to the generated Beyla configuration.
Fix	`kubeletResource.extraDiscoveryRules` and `kubeletProbes.extraDiscoveryRules` are no longer dropped.
Fix	Removed empty `image` and `container` labels from cAdvisor `container_pressure_*` metrics.
Fix	Fixed the `unrecognized block name "rule_namespace"` error when using `rules.namespaces` on the Prometheus destination.

4.1.x

Covers releases 4.1.0 through 4.1.7.

Type	Description
Breaking	Database Observability: The default `job` label is now `integrations/db-o11y` (was `integration/mysql` or `integration/postgresql`), and the `instance` label on logs is the instance name instead of the raw DSN, which moved to a `dsn` label.
Feature	Added the Kubernetes manifests feature, which captures Kubernetes manifests as log data.
Feature	Added `small`, `medium`, `large`, and `xlarge` collector presets that set Alloy resource requests and limits for common Cluster sizes.
Feature	Added `cluster.nameFrom` to set the cluster name from arbitrary sources.
Feature	Added `global.namespaceOverride` to deploy namespace-scoped resources into a custom namespace across the chart and its feature subcharts.
Feature	Added `clusterEvents.clustering`, which lets Cluster events run on a clustered collector with multiple replicas.
Feature	Added the ability to replace the contents of an Alloy component.
Feature	Added the write-ahead log (WAL) setting on the Loki destination.
Feature	Database Observability: Added `databaseObservability.excludeCurrentUser` for PostgreSQL. The per-collector `querySamples.excludeCurrentUser` is deprecated but still honored.
Enhancement	Database Observability: Added `databaseObservability.cloudProvider.gcp` for GCP Cloud SQL metadata, `sampleMinDuration` and `waitEventMinDuration` for MySQL query samples, and realigned collector defaults with Alloy.
Enhancement	Added `scrapeNativeHistograms` for the Prometheus Operator objects feature, and `convertClassicHistogramsToNhcb` to `global`.
Enhancement	Added the `timeout` setting on the OTLP destination.
Enhancement	Added the `openTelemetryConversion.keepIdentifyingResourceAttributes` option to preserve `service.name`, `service.namespace`, and `service.instance.id` on `target_info`.
Enhancement	Added `drop` policy support for tail sampling.
Enhancement	Added global pull policy support for the hooks and collectors.
Enhancement	Rendering fails with a clear error when `profiling.enabled: true` but none of the eBPF, Java, or pprof methods are enabled, instead of silently collecting nothing.
Enhancement	`labelMatchers` is now required when connecting cluster, host, or cost metrics to an existing service not deployed through `telemetryServices`. A namespace alone previously matched every Pod in the namespace.
Enhancement	Added best-effort validation that confirms `labelMatchers` select running Pods for an existing service, and that detects a port conflict with an existing Node Exporter. Both are skipped during `helm template` and `--dry-run`.
Enhancement	Platform detection now recognizes Azure AKS, Google GKE, and Amazon EKS in addition to OpenShift. The result is reported in the `platform` label of `grafana_kubernetes_monitoring_build_info` and the `provider` remote configuration attribute.
Enhancement	Updated Alloy Operator to 0.5.11, Beyla to 1.16.8, kube-state-metrics to 7.5.1, OpenCost to 2.5.25, Windows Exporter to 0.12.7, and k8s-manifest-tail to 0.1.5.
Fix	The OTLP destination `timeout` setting renders in the correct place for both the HTTP and gRPC protocols. Alloy 1.16.3 and later rejected the previous placement, which stopped the collector from starting.
Fix	`global.scrapeNativeHistograms: true` no longer crash-loops the metrics collector. The chart now adds `PrometheusProto` to the start of `scrape_protocols` when native histograms are enabled.
Fix	Service integrations render their modules when `integrations.collector` is left at its default. Previously, enabling an integration without also setting the collector produced nothing.
Fix	Alloy collector instances no longer cross-contaminate each other’s nested values, such as `mounts.extra`.
Fix	Custom destinations with `traces.enabled: true` or `profiles.enabled: true` render instead of failing with a nil pointer error.
Fix	OpenCost no longer crashes on GKE, and deploys when `customPricing` is enabled.
Fix	Beyla on OpenShift: The `SecurityContextConstraints` now sets `allowHostPorts: true`, so Beyla Pods pass admission when host networking is enabled.
Fix	The pre-delete and post-install hook templates render when `alloy-operator.waitForAlloyRemoval.securityContext` is `null`.
Fix	`extraLogProcessingStages` renders under older Helm versions, such as those bundled with the Terraform Helm provider.
Fix	Removed scrape protocols from the `prometheus.operator.*` components, which don’t support them.
Fix	`spanMetrics.excludeDimensions` values are quoted as strings, so Alloy can parse the rendered configuration.

4.0.x

Covers releases 4.0.1 through 4.0.5.

Type	Description
Feature	Added the ability to set per-signal paths for `otlp/http` destinations.
Enhancement	Added the ability to set `nodeLabels` in the host metrics feature.
Enhancement	Added an `enabled` flag to collectors.
Enhancement	PostgreSQL integration: Added the `statementsLimit` option.
Enhancement	Added validation for misplaced deployment-level settings and for clustering when required. Collector and destination names are also validated.
Enhancement	Set `appProtocol` on the Application Observability ports.
Enhancement	Cluster labels use `external_labels` instead of `write_relabel_config`.
Enhancement	Improved handling of logs from the Grafana, Loki, and Mimir integrations.
Enhancement	Updated Alloy Operator to 0.5.5, Node Exporter to 4.55.0, Beyla, kube-state-metrics, OpenCost, and Windows Exporter.
Fix	Fixed Node log level parsing, which was broken since 4.0.0.
Fix	Service discovery uses EndpointSlice instead of the deprecated Endpoints API.
Fix	Beyla discovery configuration uses glob syntax instead of regular expressions.
Fix	Fixed `serviceGraphMetrics` routing metrics to the wrong OTLP destination.
Fix	Fixed annotations and labels on Alloy custom resources.
Fix	The pre-delete hook no longer fails when the Alloy finalizer isn’t present on the resource.
Fix	Removed the `/var/configs` volume from OpenCost, which could cause problems.

4.0.0

Type	Description
Breaking	Rebuilt log features to remove `labelsToKeep` and split Loki and OpenTelemetry log gathering.
Breaking	Introduced `collectors` as a map and removed named Alloy instances.
Breaking	Converted destinations into a map.
Breaking	Removed Prometheus Operator Object CRDs.
Breaking	Extracted supplemental telemetry services from config features into their own subchart.

3.8.5

Type	Description
Enhancement	Updated Beyla, Node Exporter, OpenCost, and Prometheus Operator CRDs.
Fix	Fixed Application Observability failing to start when only traces are enabled and metrics are disabled.

3.8.4

Type	Description
Enhancement	Updated Alloy Operator to 1.5.2.
Enhancement	Updated options for `secretFilters` in Pod logs features to match changes in Alloy.
Fix	Fixed `tls_config` block name to `tls` in `otelcol.auth.oauth2` for OTLP destinations.

3.8.3

Type	Description
Feature	PostgreSQL: Added `stat_statements` `exclude_databases`, `exclude_users`, and `limit` options.
Enhancement	Disabled the OpenCost MCP server.
Enhancement	Updated Alloy Operator to 1.5.1.
Enhancement	Added more settings for profiling features.

3.8.2

Type	Description
Feature	Added TLS configuration support for OAuth2 token endpoints across all destinations.
Enhancement	Added label selectors and a completed Job filter to the Istio integration sidecar scraper and wired namespace/label selectors for Istiod discovery.
Enhancement	Updated Beyla auto-instrumentation config to use `instrument`/`exclude_instrument`, replacing deprecated `services`/`exclude_services`.
Enhancement	Updated Beyla to 1.13.0.
Enhancement	MySQL: Added `perf_schema.eventsstatements` collector options.
Enhancement	Updated Node Exporter, OpenCost, and Alloy Operator.
Enhancement	Updated `database_observability` setup for Alloy 1.13.x features.
Fix	Fixed finding Pod logs for static pods using the `config.mirror` annotation.
Fix	Fixed cAdvisor `includeNamespaces` filter dropping non-namespaced metrics like `machine_*`.
Fix	Fixed preserving user `extraEnv` in service graph collector.

3.8.1

Type	Description
Feature	Added an option to set the semantic convention version for Application Observability span names.
Fix	Fixed Java profiling so annotation targeting no longer scrapes unannotated pods.
Fix	Fixed PostgreSQL integration `includeQuery` config camelCase issue.

3.8.0

Type	Description
Feature	Added the ability to enrich metrics with pod or namespace labels.
Feature	Added the ability to set `protobufMessage` and a shortcut for the remote_write protocol.
Feature	Added a feature for gathering logs using PodLogs objects.
Enhancement	Set CRI as default logs processor if runtime is unset.
Enhancement	Automatically set required environment variables when enabling remote config.
Enhancement	Updated Windows Exporter to 0.12.3, Beyla to 1.11.0, and Prometheus Operator Object CRDs to 26.0.1.

3.7.5

Type	Description
Feature	Added configuration to specify the `overrides` section of the Span Logs Connector component for the Application Observability chart.
Enhancement	Updated Alloy Operator to 0.4.1.

3.7.4

Type	Description
Enhancement	Improved OTLP destination protocol validation.
Enhancement	Updated Prometheus Operator Object CRDs.
Fix	Fixed indentation issues for Beyla relabel.
Fix	Added `erofs` to node-exporter filesystem exclusions.

3.7.3

Type	Description
Feature	Added eBPF sample rate for profiling.
Feature	Added the ability to skip cluster metrics ServiceMonitor checks.
Enhancement	Updated Node Exporter.
Enhancement	Made the `loki.process` CRI stage `maxPartialLines` configurable.

3.7.2

Type	Description
Feature	Added `cloudProvider` configuration support to MySQL and PostgreSQL `databaseObservability`.
Enhancement	Updated Node Exporter and Alloy Operator.

3.7.1

Type	Description
Enhancement	Updated Node Exporter and kube-state-metrics.
Fix	Fixed MySQL and PostgreSQL integrations for missing log destinations and secrets.

3.7.0

Type	Description
Feature	Added an integration for Istio sidecar and service metrics.
Feature	Added an integration for PostgreSQL, including support for Database Observability.
Feature	Added the ability to set `otel_annotations` flag for the `k8sattributes` processor.
Feature	Added more options to the `secretFilter` component in the pod logs features.
Enhancement	Deployed `beyla-k8s-cache` with 1 replica by default in auto-instrumentation feature chart.
Enhancement	Updated Alloy Operator and Beyla.
Enhancement	Added check for the presence of kube-state-metrics or Node Exporter ServiceMonitors if `clusterMetrics` and `prometheusOperatorObjects` features are enabled.

3.6.2

Type	Description
Enhancement	Updated Alloy Operator to 0.3.14.
Fix	Fixed extra quotes for the sending queue storage in OTLP destinations.
Fix	Fixed the inclusion of the destination secret for Service Graph instance.

3.6.1

Type	Description
Feature	Added the ability to override the security context for the `waitForAlloyRemoval` Helm Hook.
Feature	Added the ability to define the sending queue for OTLP destinations.
Feature	Added the ability to define the remote timeout for Loki destinations.

3.6.0

Type	Description
Feature	Added Pod Init container metrics to the kube-state-metrics allow list.
Feature	Added Database Observability to the MySQL integration.
Feature	Added the ability to disable sending traces from Beyla to the Application Observability feature.
Feature	Added the ability to use ScrapeConfig objects with the Prometheus Operator Objects feature.
Fix	Updated OpenCost to include an emptydir volume mount for its config path.

3.5.7

Type	Description
Feature	Added the ability to set `scrape_native_histograms` for Prometheus scrape configs.
Enhancement	Updated kube-state-metrics and Prometheus Operator CRDs.
Fix	Fixed port assignment on the etcd integration and several cluster metrics control plane services.
Fix	Fixed MySQL integration.

3.5.6

Type	Description
Feature	Added the ability to set resources for the Helm hooks.
Feature	Added the ability to set K8s Attributes Processor filters.
Enhancement	Updated Alloy Operator and Node Exporter.

3.5.5

Type	Description
Enhancement	Updated Beyla chart to 1.9.9.
Fix	Set node label for Tempo service integration.

3.5.4

Type	Description
Enhancement	Updated Alloy Operator, Prometheus Operator CRDs, Beyla, and OpenCost.
Enhancement	Updated default Service Graph dimensions.
Fix	Ensured Jaeger Compact and Jaeger Thrift protocols use UDP.
Fix	Ensured Service Graph utilizes destination defaults.
Fix	Fixed integration scrape intervals and timeouts to check properly inside the metrics object.
Fix	Set node label for certain service integrations.

3.5.3

Type	Description
Feature	Made Alloy Profiles deployable on all nodes.
Feature	Added label selectors for the Pod Logs feature.
Feature	Added labels and annotations to Alloy CR instances.
Enhancement	Updated Alloy Operator, Beyla, OpenCost, and Prometheus Operator CRDs.

3.5.2

Type	Description
Feature	Added the ability to change pod associations.

3.5.1

Type	Description
Feature	Added the ability to set the MySQL protocol.
Enhancement	Updated `prometheus-node-exporter` from 4.47.3 to 4.48.0.
Fix	Fixed OpenCost validation when using `destinationsMap`.

3.5.0

Type	Description
Feature	Created a separate feature for Pod Logs via Kubernetes API.
Feature	Added Custom type destinations.
Feature	Added the ability to set scrape timeout everywhere.
Enhancement	Updated default resource attributes remove list and labels set for profiling feature.
Enhancement	Span metrics: Added spans prefilter for internal.

3.4.1

Type	Description
Feature	Added the ability to set labels and annotations on the hook pod.
Enhancement	Updated kube-state-metrics and OpenCost.
Enhancement	Updated Node Exporter filesystem exclusion list.
Fix	Fixed the `excludeNamespaces` option in the Prometheus Operator Object feature.

3.4.0

Type	Description
Feature	Added default attribute remove list support for OTLP destinations.
Feature	Added the ability to override attributes and set TLS settings for remote config.
Fix	Properly truncated tail sampling and service grapher Alloy instances.
Fix	Removed the “wait for Alloy operator” hook.

3.3.2

Type	Description
Feature	Added the ability to set the default scrape timeout for Prometheus Operator Objects feature.
Feature	Added `sys.env` functions to `remoteConfig` collector.
Enhancement	Updated OpenCost to 2.2.2 and Prometheus Operator CRDs to 23.0.0.
Fix	Properly truncated tail sampling and service grapher Alloy instances.

3.3.1

Type	Description
Feature	Added image pull secrets and pull policy to Helm hooks.
Enhancement	Updated OpenCost to 2.2.
Enhancement	Added transform support for span metrics.
Fix	Fixed `scrapeProtocols` typo.
Fix	Fixed hook arm64 compatibility and prevented failure if Alloy instance already deleted.

3.3.0

Type	Description
Feature	Added passthrough for `k8sattributes` processor.
Feature	Added hooks to add finalizers and Alloy instance cleanup to prevent orphaned resources.
Feature	Added pre-install hook to wait for Alloy Operator readiness before creating Alloy resources.
Feature	Added `alloy-operator.waitForReadiness` configuration option to control the timing behavior.
Feature	Allowed passing destinations as a map.
Feature	Allowed modifying Alloy settings in one place.
Feature	Added span attribute support for skip metrics generation.
Enhancement	Improved Windows Exporter discovery rules.
Enhancement	Added `kube_cronjob.*` to kube-state-metrics default allow list.

3.2.6

Type	Description
Feature	Added the ability to set error mode for filter and transform processors.
Enhancement	Updated Beyla Helm chart to 1.9.2 in auto-instrumentation feature chart.
Fix	Skipped Beyla-generated traces in span metrics.

3.2.5

Type	Description
Enhancement	Updated kube-state-metrics to 6.1.4.
Fix	Allowed destination and remote config usernames to be numbers.

3.2.4

Type	Description
Enhancement	Updated Beyla, Alloy, Alloy Operator, kube-state-metrics, and Prometheus Operator CRDs.

3.2.3

Type	Description
Enhancement	Updated Windows Exporter to 0.12.1.
Enhancement	Updated annotation handling for profiling targets.
Enhancement	Included `collector.id` in span metrics transform.
Fix	Fixed bad double quote in CRD validation.

3.2.2

Type	Description
Enhancement	Updated Alloy Operator to 0.3.7.
Enhancement	Improved how integration logs and metrics are enabled so they can be properly disabled.

3.2.1

Type	Description
Enhancement	Updated Node Exporter to 4.47.3.
Fix	Fixed newline handling at the end of the Prometheus file.
Fix	Removed check preventing deploying if `/var/log` or `/var/lib/docker/containers` are set when they may not be needed.

3.2.0

Type	Description
Feature	New feature: Profiles Receiver.
Enhancement	Prevented Node Exporter from generating metrics about `ramfs` and `tmpfs`.
Enhancement	Set the `job` label on sources from the Annotation Autodiscovery feature to more reasonable values.
Enhancement	Set `service.namespace` and `service.instance.id` labels from typical sources when using the Pod Logs feature.

Was this page helpful?

Email [email protected]

Help and support

Community