Fix spire-server PodMonitor controller-manager port names (#932)

The PodMonitor targeted the port name prom-cm, but controller-manager
containers expose pm-cm (and auto-suffixed/overridable names for external
managers), so their metrics were never scraped. Resolve port names through
a shared helper and enumerate every controller-manager in the PodMonitor.

Signed-off-by: sabsari <[email protected]>
Co-authored-by: Claude Opus 4.8 <[email protected]>
This commit is contained in:
sabsari
2026-08-30 08:44:56 -07:00
committed by GitHub
co-authored by Claude Opus 4.8
parent cb0f08e53a
commit 85ab0fa5ab
3 changed files with 174 additions and 27 deletions
@@ -1,3 +1,72 @@
{{/* 15-char-safe port-name suffix for a controller-manager cluster name ("" -> ""). Shared by container ports and the PodMonitor so they cannot drift. */}}
{{- define "spire-controller-manager.portSuffix" -}}
{{- $name := . -}}
{{- $portSuffix := "" -}}
{{- if ne $name "" -}}
{{- $portSuffix = printf "-%s" $name -}}
{{- if gt (len $name) 9 -}}
{{- $numberMatch := regexFind "[-]?[0-9]{1,2}$" $name -}}
{{- if $numberMatch -}}
{{- $numLen := len $numberMatch -}}
{{- $baseLen := sub (len $name) $numLen | int -}}
{{- $baseName := substr 0 $baseLen $name -}}
{{- if not (hasPrefix "-" $numberMatch) -}}
{{- $numberMatch = printf "-%s" $numberMatch -}}
{{- end -}}
{{- $maxBase := sub 9 (len $numberMatch) | int -}}
{{- $baseName = $baseName | trunc $maxBase | trimSuffix "-" -}}
{{- $portSuffix = printf "-%s%s" $baseName $numberMatch -}}
{{- else -}}
{{- $hash := sha256sum $name | trunc 3 -}}
{{- $portSuffix = printf "-%s-%s" ($name | trunc 5 | trimSuffix "-") $hash -}}
{{- end -}}
{{- end -}}
{{- end -}}
{{- $portSuffix -}}
{{- end -}}
{{/* Resolve a port name: override wins, else prefix+suffix. dict: prefix, portSuffix, override */}}
{{- define "spire-controller-manager.portName" -}}
{{- if and (hasKey . "override") (ne (.override | toString) "") -}}
{{- .override -}}
{{- else -}}
{{- printf "%s%s" .prefix .portSuffix -}}
{{- end -}}
{{- end -}}
{{/* Prometheus port name for one controller-manager. dict: name (""=main), settings (may hold prometheusPortName) */}}
{{- define "spire-controller-manager.promPortName" -}}
{{- $override := "" -}}
{{- if hasKey .settings "prometheusPortName" -}}
{{- $override = .settings.prometheusPortName -}}
{{- end -}}
{{- include "spire-controller-manager.portName" (dict "prefix" "pm-cm" "portSuffix" (include "spire-controller-manager.portSuffix" .name) "override" $override) -}}
{{- end -}}
{{/* List of prometheus port names for every controller-manager that renders one. Consumed by the PodMonitor. */}}
{{- define "spire-controller-manager.prometheusPortNames" -}}
{{- $root := . -}}
{{- $names := list -}}
{{- if eq (.Values.controllerManager.enabled | toString) "true" -}}
{{- $names = append $names (include "spire-controller-manager.promPortName" (dict "name" "" "settings" .Values.controllerManager)) -}}
{{- end -}}
{{- if .Values.externalControllerManagers.enabled -}}
{{- $clusters := default .Values.kubeConfigs .Values.externalControllerManagers.clusters -}}
{{- range $name, $_ := $clusters -}}
{{- $clusterSettings := dict -}}
{{- if hasKey $root.Values.externalControllerManagers.clusters $name -}}
{{- $clusterSettings = index $root.Values.externalControllerManagers.clusters $name -}}
{{- end -}}
{{- $pmName := include "spire-controller-manager.promPortName" (dict "name" $name "settings" $clusterSettings) -}}
{{- if has $pmName $names -}}
{{- fail (printf "controller-manager prometheus port name %q collides for cluster %q; set a distinct prometheusPortName override" $pmName $name) -}}
{{- end -}}
{{- $names = append $names $pmName -}}
{{- end -}}
{{- end -}}
{{- $names | toYaml -}}
{{- end -}}
{{- define "spire-controller-manager.containers" }}
{{- $root := . }}
{{- $settings := dict }}
@@ -41,23 +110,7 @@ Auto-generation preserves trailing numbers from cluster names or uses hash for u
{{- $prometheusPortName = $clusterSettings.prometheusPortName }}
{{- end }}
{{- if or (eq $healthPortName "") (eq $prometheusPortName "") }}
{{- if gt (len $name) 9 }}
{{- $numberMatch := regexFind "[-]?[0-9]{1,2}$" $name }}
{{- if $numberMatch }}
{{- $numLen := len $numberMatch }}
{{- $baseLen := sub (len $name) $numLen | int }}
{{- $baseName := substr 0 $baseLen $name }}
{{- if not (hasPrefix "-" $numberMatch) }}
{{- $numberMatch = printf "-%s" $numberMatch }}
{{- end }}
{{- $maxBase := sub 9 (len $numberMatch) | int }}
{{- $baseName = $baseName | trunc $maxBase | trimSuffix "-" }}
{{- $portSuffix = printf "-%s%s" $baseName $numberMatch }}
{{- else }}
{{- $hash := sha256sum $name | trunc 3 }}
{{- $portSuffix = printf "-%s-%s" ($name | trunc 5 | trimSuffix "-") $hash }}
{{- end }}
{{- end }}
{{- $portSuffix = include "spire-controller-manager.portSuffix" $name }}
{{- end }}
{{- $startPort = add $startPort 2 }}
@@ -127,17 +180,11 @@ Auto-generation preserves trailing numbers from cluster names or uses hash for u
containerPort: 9443
protocol: TCP
{{- end }}
{{- $hpName := .healthPortName }}
{{- if eq $hpName "" }}
{{- $hpName = printf "hp-cm%s" .portSuffix }}
{{- end }}
{{- $hpName := include "spire-controller-manager.portName" (dict "prefix" "hp-cm" "portSuffix" .portSuffix "override" .healthPortName) }}
- containerPort: {{ $healthPort }}
name: {{ $hpName }}
{{- if or (dig "telemetry" "prometheus" "enabled" .Values.telemetry.prometheus.enabled .Values.global) (and (dig "spire" "recommendations" "enabled" false .Values.global) (dig "spire" "recommendations" "prometheus" true .Values.global)) }}
{{- $pmName := .prometheusPortName }}
{{- if eq $pmName "" }}
{{- $pmName = printf "pm-cm%s" .portSuffix }}
{{- end }}
{{- $pmName := include "spire-controller-manager.portName" (dict "prefix" "pm-cm" "portSuffix" .portSuffix "override" .prometheusPortName) }}
- containerPort: {{ $promPort }}
name: {{ $pmName }}
{{- end }}
@@ -21,7 +21,9 @@ spec:
{{- include "spire-server.selectorLabels" . | nindent 6 }}
podMetricsEndpoints:
- port: prom
- port: prom-cm
{{- range (include "spire-controller-manager.prometheusPortNames" . | fromYamlArray) }}
- port: {{ . }}
{{- end }}
{{- if ne $namespace $podNamespace }}
namespaceSelector:
kubernetes.io/metadata.name: {{ $podNamespace }}
+98
View File
@@ -810,6 +810,104 @@ spire-server:
}
})
})
Describe("spire-server.telemetry.podMonitor controller-manager ports", func() {
podMonitorTmpl := "spire/charts/spire-server/templates/podmonitor.yaml"
serverTmpl := "spire/charts/spire-server/templates/server-resource.yaml"
It("targets the real main controller-manager port name, not the legacy prom-cm", func() {
objs, err := ValueStringRender(chart, `
spire-server:
controllerManager:
enabled: true
telemetry:
prometheus:
enabled: true
podMonitor:
enabled: true
`)
Expect(err).Should(Succeed())
podMonitor := objs[podMonitorTmpl]
Expect(podMonitor).Should(ContainSubstring("- port: prom"))
Expect(podMonitor).Should(ContainSubstring("- port: pm-cm"))
// Regression: the PodMonitor used to hardcode a port name the container never renders.
Expect(podMonitor).ShouldNot(ContainSubstring("prom-cm"))
// The endpoint must match the actual container port.
Expect(objs[serverTmpl]).Should(ContainSubstring("name: pm-cm"))
})
It("enumerates external controller-managers, honouring auto-suffix and prometheusPortName override", func() {
objs, err := ValueStringRender(chart, `
spire-server:
controllerManager:
enabled: true
telemetry:
prometheus:
enabled: true
podMonitor:
enabled: true
kubeConfigs:
child01:
kubeConfig: |
apiVersion: v1
kind: Config
verylongclustername:
kubeConfig: |
apiVersion: v1
kind: Config
externalControllerManagers:
enabled: true
clusters:
child01:
kubeConfigName: child01
verylongclustername:
kubeConfigName: verylongclustername
prometheusPortName: prom-ext2
`)
Expect(err).Should(Succeed())
podMonitor := objs[podMonitorTmpl]
server := objs[serverTmpl]
// Auto-suffixed external CM and the overridden one must both be scraped,
// and each endpoint must match a real container port name.
for _, port := range []string{"pm-cm", "pm-cm-child01", "prom-ext2"} {
Expect(podMonitor).Should(ContainSubstring("- port: " + port))
Expect(server).Should(ContainSubstring("name: " + port))
}
})
It("derives external controller-managers from kubeConfigs when clusters is the default {}", func() {
objs, err := ValueStringRender(chart, `
spire-server:
controllerManager:
enabled: true
telemetry:
prometheus:
enabled: true
podMonitor:
enabled: true
kubeConfigs:
child01:
kubeConfig: |
apiVersion: v1
kind: Config
child02:
kubeConfig: |
apiVersion: v1
kind: Config
externalControllerManagers:
enabled: true
clusters: {}
`)
Expect(err).Should(Succeed())
podMonitor := objs[podMonitorTmpl]
server := objs[serverTmpl]
// clusters={} (the chart default) falls back to kubeConfigs, so each
// kubeConfig-derived controller-manager must be scraped and match its port.
for _, port := range []string{"pm-cm", "pm-cm-child01", "pm-cm-child02"} {
Expect(podMonitor).Should(ContainSubstring("- port: " + port))
Expect(server).Should(ContainSubstring("name: " + port))
}
})
})
Describe("spire-server.dataStore.sql.postgres passwordless", func() {
It("omits password and the -dbpw Secret for cert auth with an empty password", func() {
objs, err := ValueStringRender(chart, `