k8s.io/kubernetes@v1.29.3/pkg/controller/nodelifecycle/metrics.go (about)

     1  /*
     2  Copyright 2017 The Kubernetes Authors.
     3  
     4  Licensed under the Apache License, Version 2.0 (the "License");
     5  you may not use this file except in compliance with the License.
     6  You may obtain a copy of the License at
     7  
     8      http://www.apache.org/licenses/LICENSE-2.0
     9  
    10  Unless required by applicable law or agreed to in writing, software
    11  distributed under the License is distributed on an "AS IS" BASIS,
    12  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    13  See the License for the specific language governing permissions and
    14  limitations under the License.
    15  */
    16  
    17  package nodelifecycle
    18  
    19  import (
    20  	"sync"
    21  
    22  	"k8s.io/component-base/metrics"
    23  	"k8s.io/component-base/metrics/legacyregistry"
    24  )
    25  
    26  const (
    27  	nodeControllerSubsystem = "node_collector"
    28  	zoneHealthStatisticKey  = "zone_health"
    29  	zoneSizeKey             = "zone_size"
    30  	zoneNoUnhealthyNodesKey = "unhealthy_nodes_in_zone"
    31  	evictionsTotalKey       = "evictions_total"
    32  
    33  	updateNodeHealthKey     = "update_node_health_duration_seconds"
    34  	updateAllNodesHealthKey = "update_all_nodes_health_duration_seconds"
    35  )
    36  
    37  var (
    38  	zoneHealth = metrics.NewGaugeVec(
    39  		&metrics.GaugeOpts{
    40  			Subsystem:      nodeControllerSubsystem,
    41  			Name:           zoneHealthStatisticKey,
    42  			Help:           "Gauge measuring percentage of healthy nodes per zone.",
    43  			StabilityLevel: metrics.ALPHA,
    44  		},
    45  		[]string{"zone"},
    46  	)
    47  	zoneSize = metrics.NewGaugeVec(
    48  		&metrics.GaugeOpts{
    49  			Subsystem:      nodeControllerSubsystem,
    50  			Name:           zoneSizeKey,
    51  			Help:           "Gauge measuring number of registered Nodes per zones.",
    52  			StabilityLevel: metrics.ALPHA,
    53  		},
    54  		[]string{"zone"},
    55  	)
    56  	unhealthyNodes = metrics.NewGaugeVec(
    57  		&metrics.GaugeOpts{
    58  			Subsystem:      nodeControllerSubsystem,
    59  			Name:           zoneNoUnhealthyNodesKey,
    60  			Help:           "Gauge measuring number of not Ready Nodes per zones.",
    61  			StabilityLevel: metrics.ALPHA,
    62  		},
    63  		[]string{"zone"},
    64  	)
    65  	evictionsTotal = metrics.NewCounterVec(
    66  		&metrics.CounterOpts{
    67  			Subsystem:      nodeControllerSubsystem,
    68  			Name:           evictionsTotalKey,
    69  			Help:           "Number of Node evictions that happened since current instance of NodeController started.",
    70  			StabilityLevel: metrics.STABLE,
    71  		},
    72  		[]string{"zone"},
    73  	)
    74  
    75  	updateNodeHealthDuration = metrics.NewHistogram(
    76  		&metrics.HistogramOpts{
    77  			Subsystem:      nodeControllerSubsystem,
    78  			Name:           updateNodeHealthKey,
    79  			Help:           "Duration in seconds for NodeController to update the health of a single node.",
    80  			Buckets:        metrics.ExponentialBuckets(0.001, 4, 8), // 1ms -> ~15s
    81  			StabilityLevel: metrics.ALPHA,
    82  		},
    83  	)
    84  	updateAllNodesHealthDuration = metrics.NewHistogram(
    85  		&metrics.HistogramOpts{
    86  			Subsystem:      nodeControllerSubsystem,
    87  			Name:           updateAllNodesHealthKey,
    88  			Help:           "Duration in seconds for NodeController to update the health of all nodes.",
    89  			Buckets:        metrics.ExponentialBuckets(0.01, 4, 8), // 10ms -> ~3m
    90  			StabilityLevel: metrics.ALPHA,
    91  		},
    92  	)
    93  )
    94  
    95  var registerMetrics sync.Once
    96  
    97  // Register the metrics that are to be monitored.
    98  func Register() {
    99  	registerMetrics.Do(func() {
   100  		legacyregistry.MustRegister(zoneHealth)
   101  		legacyregistry.MustRegister(zoneSize)
   102  		legacyregistry.MustRegister(unhealthyNodes)
   103  		legacyregistry.MustRegister(evictionsTotal)
   104  		legacyregistry.MustRegister(updateNodeHealthDuration)
   105  		legacyregistry.MustRegister(updateAllNodesHealthDuration)
   106  	})
   107  }