If you use JENNIFER's hexagonal pod view to check the pods on each node, the switch to bars can take some getting used to. At 25 worker nodes or more, the Cluster Dashboard uses a compact bar view to fit the nodes on the screen. It still gives you an overview, but the individual pod view may feel easier to read.

A user-defined dashboard offers a workaround: add multiple Cluster Map charts and select a different group of nodes for each one. With fewer than 25 nodes in each chart, you can keep the hexagonal view and see all the groups on the same dashboard.

Side-by-side comparison of the hexagonal pod view with 15 worker nodes and the compact bar view with 30 worker nodes.
Figure 1. The hexagonal pod view on the left and the compact bar view on the right. These examples show 15 and 30 worker nodes, respectively. The display changes at 25 or more worker nodes.

Split 30 worker nodes across two maps

For a cluster with 30 worker nodes, select worker-01 through worker-15 in the first Cluster Map and worker-16 through worker-30 in the second. Set both charts to the same cluster.

Two Cluster Maps for the same cluster. The left map contains worker-01 through worker-15; the right contains worker-16 through worker-30. Each node contains a regular grid of hexagonal pods.
Figure 2. The same cluster split into two groups of 15 worker nodes. This is an illustrative layout, not a product screenshot. The numbers beside the pod groups are example pod counts; the status colors are also illustrative.

Each chart now has 15 nodes, so both can use the hexagonal pod view. Place them side by side to keep all 30 worker nodes in view. The two circular maps show different parts of one cluster. The cluster and pod placement stay as they are; only the charts' display scope changes.

SettingCluster Map ACluster Map B
Target clusterThe same clusterThe same cluster
Selected nodesworker-01 through worker-15worker-16 through worker-30
Node count1515

Filter each chart by node

Use Manual Selection in the Cluster Map settings to assign nodes to each chart. This fixes the chart's target scope: changing the cluster or namespace selector at the top of the dashboard will not change the chart's selected targets.

Both charts start with the same list of nodes. In Chart A, check worker-01 through worker-15 and leave worker-16 through worker-30 unchecked. Reverse the selection in Chart B. The checked nodes determine which part of the cluster each map shows.

Two illustrated node selection panels for the same cluster. Both list all 30 worker nodes. Chart A checks nodes 01 through 15 and leaves 16 through 30 unchecked. Chart B checks nodes 16 through 30 and leaves 01 through 15 unchecked.
Figure 3. Select a different half of the same node list in each chart. Each chart has 15 checked nodes, with no overlap between the groups. This settings illustration uses example node names and is not a product screenshot.
  1. Add a Cluster Map chart to a user-defined dashboard.
  2. Under Target Selection, choose Manual Selection. Set Target cluster to the cluster you want to monitor.
  3. Under Resource, choose Node. Check worker-01 through worker-15, leave the other nodes unchecked, and save the chart.
  4. Add a second Cluster Map for the same cluster. Check worker-16 through worker-30 and leave worker-01 through worker-15 unchecked, then save that chart.
  5. Arrange the charts side by side and give each enough space to make the pods readable. Use titles that identify the cluster and node group, then save the dashboard.

Keep the node lists up to date

Before using the dashboard, check that the two selections cover every worker node you want to monitor, with no unintended duplicates. Check the selections again when nodes are added or replaced so that new nodes are included. If your monitoring scope also covers pods running on control plane nodes, include those nodes too.

The group size does not have to be 15. Choose groups that suit your node count and screen space, keeping each chart below 25 nodes.

The figures are explanatory illustrations based on JENNIFER's cluster visualization and chart settings. Node names, pod counts, and status colors are examples.