Merge pull request ceph#60768 from VallariAg/wip-prometheus-tested-limits

monitoring: Add alert NVMeoFTooManyNamespaces

Reviewed-by: Afreen Misbah <[email protected]>

Author: afreen23

monitoring/ceph-mixin/config.libsonnet

@@ -12,7 +12,8 @@
     NVMeoFMaxGatewaysPerGroup: 4,
     NVMeoFMaxGatewaysPerCluster: 4,
     NVMeoFHighGatewayCPU: 80,
-    NVMeoFMaxSubsystemsPerGateway: 16,
+    NVMeoFMaxSubsystemsPerGateway: 128,
+    NVMeoFMaxNamespaces: 1024,
     NVMeoFHighClientCount: 32,
     NVMeoFHighHostCPU: 80,
     //

monitoring/ceph-mixin/prometheus_alerts.libsonnet

@@ -908,13 +908,23 @@
         {
           alert: 'NVMeoFTooManySubsystems',
           'for': '1m',
-          expr: 'count by(gateway_host, cluster) (label_replace(ceph_nvmeof_subsystem_metadata,"gateway_host","$1","instance","(.*):.*")) > %.2f' % [$._config.NVMeoFMaxSubsystemsPerGateway],
+          expr: 'count by(gateway_host, cluster) (label_replace(ceph_nvmeof_subsystem_metadata,"gateway_host","$1","instance","(.*?)(?::.*)?")) > %.2f' % [$._config.NVMeoFMaxSubsystemsPerGateway],
           labels: { severity: 'warning', type: 'ceph_default' },
           annotations: {
             summary: 'The number of subsystems defined to the gateway exceeds supported values%(cluster)s' % $.MultiClusterSummary(),
             description: 'Although you may continue to create subsystems in {{ $labels.gateway_host }}, the configuration may not be supported',
           },
         },
+        {
+          alert: 'NVMeoFTooManyNamespaces',
+          'for': '1m',
+          expr: 'sum by(gateway_host, cluster) (label_replace(ceph_nvmeof_subsystem_namespace_count,"gateway_host","$1","instance","(.*?)(?::.*)?")) > %.2f' % [$._config.NVMeoFMaxNamespaces],
+          labels: { severity: 'warning', type: 'ceph_default' },
+          annotations: {
+            summary: 'The number of namespaces defined to the gateway exceeds supported values%(cluster)s' % $.MultiClusterSummary(),
+            description: 'Although you may continue to create namespaces in {{ $labels.gateway_host }}, the configuration may not be supported',
+          },
+        },
         {
           alert: 'NVMeoFVersionMismatch',
           'for': '1h',

monitoring/ceph-mixin/prometheus_alerts.yml

@@ -814,7 +814,16 @@ groups:
         annotations:
           description: "Although you may continue to create subsystems in {{ $labels.gateway_host }}, the configuration may not be supported"
           summary: "The number of subsystems defined to the gateway exceeds supported values on cluster {{ $labels.cluster }}"
-        expr: "count by(gateway_host, cluster) (label_replace(ceph_nvmeof_subsystem_metadata,\"gateway_host\",\"$1\",\"instance\",\"(.*):.*\")) > 16.00"
+        expr: "count by(gateway_host, cluster) (label_replace(ceph_nvmeof_subsystem_metadata,\"gateway_host\",\"$1\",\"instance\",\"(.*?)(?::.*)?\")) > 128.00"
+        for: "1m"
+        labels:
+          severity: "warning"
+          type: "ceph_default"
+      - alert: "NVMeoFTooManyNamespaces"
+        annotations:
+          description: "Although you may continue to create namespaces in {{ $labels.gateway_host }}, the configuration may not be supported"
+          summary: "The number of namespaces defined to the gateway exceeds supported values on cluster {{ $labels.cluster }}"
+        expr: "sum by(gateway_host, cluster) (label_replace(ceph_nvmeof_subsystem_namespace_count,\"gateway_host\",\"$1\",\"instance\",\"(.*?)(?::.*)?\")) > 1024.00"
         for: "1m"
         labels:
           severity: "warning"