summary.go 4.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151
  1. /*
  2. Copyright 2016 The Kubernetes Authors.
  3. Licensed under the Apache License, Version 2.0 (the "License");
  4. you may not use this file except in compliance with the License.
  5. You may obtain a copy of the License at
  6. http://www.apache.org/licenses/LICENSE-2.0
  7. Unless required by applicable law or agreed to in writing, software
  8. distributed under the License is distributed on an "AS IS" BASIS,
  9. WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  10. See the License for the specific language governing permissions and
  11. limitations under the License.
  12. */
  13. package stats
  14. import (
  15. "fmt"
  16. "k8s.io/klog"
  17. metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
  18. statsapi "k8s.io/kubernetes/pkg/kubelet/apis/stats/v1alpha1"
  19. "k8s.io/kubernetes/pkg/kubelet/util"
  20. )
  21. // SummaryProvider provides summaries of the stats from Kubelet.
  22. type SummaryProvider interface {
  23. // Get provides a new Summary with the stats from Kubelet,
  24. // and will update some stats if updateStats is true
  25. Get(updateStats bool) (*statsapi.Summary, error)
  26. // GetCPUAndMemoryStats provides a new Summary with the CPU and memory stats from Kubelet,
  27. GetCPUAndMemoryStats() (*statsapi.Summary, error)
  28. }
  29. // summaryProviderImpl implements the SummaryProvider interface.
  30. type summaryProviderImpl struct {
  31. // kubeletCreationTime is the time at which the summaryProvider was created.
  32. kubeletCreationTime metav1.Time
  33. // systemBootTime is the time at which the system was started
  34. systemBootTime metav1.Time
  35. provider Provider
  36. }
  37. var _ SummaryProvider = &summaryProviderImpl{}
  38. // NewSummaryProvider returns a SummaryProvider using the stats provided by the
  39. // specified statsProvider.
  40. func NewSummaryProvider(statsProvider Provider) SummaryProvider {
  41. kubeletCreationTime := metav1.Now()
  42. bootTime, err := util.GetBootTime()
  43. if err != nil {
  44. // bootTime will be zero if we encounter an error getting the boot time.
  45. klog.Warningf("Error getting system boot time. Node metrics will have an incorrect start time: %v", err)
  46. }
  47. return &summaryProviderImpl{
  48. kubeletCreationTime: kubeletCreationTime,
  49. systemBootTime: metav1.NewTime(bootTime),
  50. provider: statsProvider,
  51. }
  52. }
  53. func (sp *summaryProviderImpl) Get(updateStats bool) (*statsapi.Summary, error) {
  54. // TODO(timstclair): Consider returning a best-effort response if any of
  55. // the following errors occur.
  56. node, err := sp.provider.GetNode()
  57. if err != nil {
  58. return nil, fmt.Errorf("failed to get node info: %v", err)
  59. }
  60. nodeConfig := sp.provider.GetNodeConfig()
  61. rootStats, networkStats, err := sp.provider.GetCgroupStats("/", updateStats)
  62. if err != nil {
  63. return nil, fmt.Errorf("failed to get root cgroup stats: %v", err)
  64. }
  65. rootFsStats, err := sp.provider.RootFsStats()
  66. if err != nil {
  67. return nil, fmt.Errorf("failed to get rootFs stats: %v", err)
  68. }
  69. imageFsStats, err := sp.provider.ImageFsStats()
  70. if err != nil {
  71. return nil, fmt.Errorf("failed to get imageFs stats: %v", err)
  72. }
  73. var podStats []statsapi.PodStats
  74. if updateStats {
  75. podStats, err = sp.provider.ListPodStatsAndUpdateCPUNanoCoreUsage()
  76. } else {
  77. podStats, err = sp.provider.ListPodStats()
  78. }
  79. if err != nil {
  80. return nil, fmt.Errorf("failed to list pod stats: %v", err)
  81. }
  82. rlimit, err := sp.provider.RlimitStats()
  83. if err != nil {
  84. return nil, fmt.Errorf("failed to get rlimit stats: %v", err)
  85. }
  86. nodeStats := statsapi.NodeStats{
  87. NodeName: node.Name,
  88. CPU: rootStats.CPU,
  89. Memory: rootStats.Memory,
  90. Network: networkStats,
  91. StartTime: sp.systemBootTime,
  92. Fs: rootFsStats,
  93. Runtime: &statsapi.RuntimeStats{ImageFs: imageFsStats},
  94. Rlimit: rlimit,
  95. SystemContainers: sp.GetSystemContainersStats(nodeConfig, podStats, updateStats),
  96. }
  97. summary := statsapi.Summary{
  98. Node: nodeStats,
  99. Pods: podStats,
  100. }
  101. return &summary, nil
  102. }
  103. func (sp *summaryProviderImpl) GetCPUAndMemoryStats() (*statsapi.Summary, error) {
  104. // TODO(timstclair): Consider returning a best-effort response if any of
  105. // the following errors occur.
  106. node, err := sp.provider.GetNode()
  107. if err != nil {
  108. return nil, fmt.Errorf("failed to get node info: %v", err)
  109. }
  110. nodeConfig := sp.provider.GetNodeConfig()
  111. rootStats, err := sp.provider.GetCgroupCPUAndMemoryStats("/", false)
  112. if err != nil {
  113. return nil, fmt.Errorf("failed to get root cgroup stats: %v", err)
  114. }
  115. podStats, err := sp.provider.ListPodCPUAndMemoryStats()
  116. if err != nil {
  117. return nil, fmt.Errorf("failed to list pod stats: %v", err)
  118. }
  119. nodeStats := statsapi.NodeStats{
  120. NodeName: node.Name,
  121. CPU: rootStats.CPU,
  122. Memory: rootStats.Memory,
  123. StartTime: rootStats.StartTime,
  124. SystemContainers: sp.GetSystemContainersCPUAndMemoryStats(nodeConfig, podStats, false),
  125. }
  126. summary := statsapi.Summary{
  127. Node: nodeStats,
  128. Pods: podStats,
  129. }
  130. return &summary, nil
  131. }