From 3e36c10349ad8d18a845b5bc2187fd94ee417c1d Mon Sep 17 00:00:00 2001
From: Erik Sundell <erik.sundell+2025@sensmetry.com>
Date: Tue, 11 Nov 2025 11:29:38 +0100
Subject: [PATCH] Make non-running pods panels consistent and style them
 intuitively

---
 dashboards/cluster.jsonnet    | 21 ++++++-------
 dashboards/common.libsonnet   | 55 +++++++++++++++++++++++++++++++++++
 dashboards/jupyterhub.jsonnet |  4 +--
 3 files changed, 67 insertions(+), 13 deletions(-)

diff --git a/dashboards/cluster.jsonnet b/dashboards/cluster.jsonnet
index d20eff1..88783f1 100755
--- a/dashboards/cluster.jsonnet
+++ b/dashboards/cluster.jsonnet
@@ -403,22 +403,23 @@ local nodeOOMKills =
   ]);
 
 local nonRunningPods =
-  common.barChartOptions
-  + barChart.new('Pods not in Running state')
-  + barChart.panelOptions.withDescription(
+  common.tsOptions
+  + common.tsPodStateStylingOverrides
+  + ts.new('Non Running Pods')
+  + ts.panelOptions.withDescription(
     |||
-      Pods in states other than 'Running'.
+      Pods in a non-running state in the hub's namespace.
 
-      In a functional clusters, pods should not be in non-Running states for long.
-    |||,
+      Pods stuck in non-running states often indicate an error condition
+    |||
   )
-  + ts.fieldConfig.defaults.custom.stacking.withMode('normal')
-  + barChart.standardOptions.withDecimals(0)
-  + barChart.queryOptions.withTargets([
+  + ts.queryOptions.withTargets([
     prometheus.new(
       '$PROMETHEUS_DS',
       |||
-        sum(kube_pod_status_phase{phase!="Running"}) by (phase)
+        sum(
+          kube_pod_status_phase{phase!="Running"}
+        ) by (phase)
       |||
     )
     + prometheus.withLegendFormat('{{phase}}'),
diff --git a/dashboards/common.libsonnet b/dashboards/common.libsonnet
index 14f60dc..dd29b98 100644
--- a/dashboards/common.libsonnet
+++ b/dashboards/common.libsonnet
@@ -93,6 +93,61 @@ local _getDashedLineOverride(pattern, color) = {
     ])
   ,
 
+  tsPodStateStylingOverrides:
+    ts.standardOptions.withOverrides([
+      {
+        matcher: { id: 'byName', options: 'Pending' },
+        properties: [{
+          id: 'color',
+          value: {
+            fixedColor: 'yellow',
+            mode: 'fixed',
+          },
+        }],
+      },
+      {
+        matcher: { id: 'byName', options: 'Running' },
+        properties: [{
+          id: 'color',
+          value: {
+            fixedColor: 'blue',
+            mode: 'fixed',
+          },
+        }],
+      },
+      {
+        matcher: { id: 'byName', options: 'Succeeded' },
+        properties: [{
+          id: 'color',
+          value: {
+            fixedColor: 'green',
+            mode: 'fixed',
+          },
+        }],
+      },
+      {
+        matcher: { id: 'byName', options: 'Unknown' },
+        properties: [{
+          id: 'color',
+          value: {
+            fixedColor: 'orange',
+            mode: 'fixed',
+          },
+        }],
+      },
+      {
+        matcher: { id: 'byName', options: 'Failed' },
+        properties: [{
+          id: 'color',
+          value: {
+            fixedColor: 'red',
+            mode: 'fixed',
+          },
+        }],
+      },
+    ])
+  ,
+
   // grafonnet ref: https://grafana.github.io/grafonnet/API/dashboard/variable.html
   variables: {
     prometheus:
diff --git a/dashboards/jupyterhub.jsonnet b/dashboards/jupyterhub.jsonnet
index 62574d1..c74b9a8 100755
--- a/dashboards/jupyterhub.jsonnet
+++ b/dashboards/jupyterhub.jsonnet
@@ -276,6 +276,7 @@ local usersPerNode =
 
 local nonRunningPods =
   common.tsOptions
+  + common.tsPodStateStylingOverrides
   + ts.new('Non Running Pods')
   + ts.panelOptions.withDescription(
     |||
@@ -284,8 +285,6 @@ local nonRunningPods =
       Pods stuck in non-running states often indicate an error condition
     |||
   )
-  // decimalsY1=0,
-  + ts.fieldConfig.defaults.custom.stacking.withMode('normal')
   + ts.queryOptions.withTargets([
     prometheus.new(
       '$PROMETHEUS_DS',
@@ -311,7 +310,6 @@ local sharedVolumeFreeSpace =
       what extra deployment is needed.
     |||
   )
-  // decimalsY1=0,
   + ts.standardOptions.withMax(1)
   + ts.standardOptions.withUnit('percentunit')
   + ts.queryOptions.withTargets([