Skip to content
This repository has been archived by the owner on Dec 8, 2023. It is now read-only.

Latest commit

 

History

History
304 lines (221 loc) · 9.26 KB

configuration.md

File metadata and controls

304 lines (221 loc) · 9.26 KB
Raw

Configuration

The Wavefront Collector for Kubernetes is configured via command-line flags and a configuration file.

Starting with version 1.0, most command line flags have been deprecated in favor of a top-level configuration file.

Flags

Usage of ./wavefront-collector:
      --config-file string             required configuration file
      --daemon                         enable daemon mode (required when running as daemonset)
      --log-level string               one of info, debug or trace (default "info")
      --profile                        enable pprof (for debugging)
      --version                        print version info and exit
      --max-procs int                  max number of CPUs that can be used simultaneously.
                                       Less than 1 for default (number of cores)

Configuration file

Source: config.go

The configuration file is written in YAML and provided using the --config-file flag. The Collector can reload configuration changes at runtime.

A reference example is provided here.

# An unique identifier for your Kubernetes cluster. Defaults to 'k8s-cluster'.
# Included as a point tag on all metrics reported to Wavefront.
clusterName: k8s-cluster

# Whether auto-discovery is enabled. Defaults to true.
enableDiscovery: true

# Whether event collection is enabled. Requires Wavefront Proxy v6.0+.
enableEvents: false

# The global interval at which data is flushed. Defaults to 60 seconds.
# Duration type specified as [0-9]+(ms|[smhdwy])
flushInterval: 60s

# The global interval at which data is collected. Defaults to 60 seconds.
# Duration type specified as [0-9]+(ms|[smhdwy])
# Note: collection intervals can be overridden per source.
defaultCollectionInterval: 60s

# Timeout for sinks to export data to Wavefront. Defaults to 20 seconds.
# Duration type specified as [0-9]+(ms|[smhdwy])
sinkExportDataTimeout: 20s

# Required: List of Wavefront sinks. At least 1 required.
sinks:
  # see the Wavefront sink section for details

sources:
  # Required: Source for collecting metrics from the kubelet stats summary API.
  kubernetes_source:
    # see kubernetes_source for details

  # Optional source for collecting metrics from cluster level Kubernetes resources.
  kubernetes_state_source:
    # see kubernetes_state_source for details
  
  # Optional source for collecting cAdvisor metrics
  kubernetes_cadvisor_source:
    # see kubernetes_cadvisor_source for details

  # Optional source for collecting control plane metrics
  kubernetes_control_plane_source:
    # see kubernetes_control_plane_source for details

  # Optional source for emitting internal collector stats.
  internal_stats_source:
    # see internal_stats_source for details

  # Optional list of prometheus sources.
  prometheus_sources:
    # see prometheus_source for details

  # Optional list of telegraf sources.
  telegraf_sources:
    # see telegraf_source for details    

  # Optional source for collecting host level systemd unit metrics.
  systemd_source:
    # see systemd_source for details

# Optional auto-discovery configuration.
discovery:
  # optional prefix for annotation based discovery.
  # when specified, this replaces prefixes such as `prometheus.io` etc
  annotation_prefix: "prometheus.io"

  # frequency of evaluating discovery rules. Defaults to 5 minutes.
  # format is [0-9]+(ms|[smhdwy])
  discovery_interval: 5m

  # enables support for sourcing runtime discovery rules from configmaps
  # with the annotation "wavefront.com/discovery-config: 'true'". Default is false.
  enable_runtime_plugins: true

  # disables discovery based on annotations. Default is false.
  disable_annotation_discovery: false

  plugins:
  # see auto-discovery for details

# Optional event collection configuration
events:
  # optional filtering of events collected
  filters:
  # see the filtering documentation for details

Wavefront sink

# The Wavefront proxy address of the form 'hostname:port'.
proxyAddress: wavefront-proxy.default.svc.cluster.local:2878

# Wavefront URL of the form https:YOUR_INSTANCE.wavefront.com. Only required for direct ingestion.
server: https://<instance>.wavefront.com

# Wavefront API token with direct data ingestion permission. Only required for direct ingestion.
token: <string>

kubernetes_source

# Defaults to empty string when using port 10255.
url: 'https://kubernetes.default.svc'

# Either 10255 (default, read-only kubelet port) or 10250 (secure kubelet port).
kubeletPort: <10250|10255>

# Defaults to false. Set to true if `kubeletPort` set to 10250.
kubeletHttps: <true|false>

# Defaults to true.
inClusterConfig: <true|false>

# Defaults to false.
useServiceAccount: <true|false>

# Defaults to false.
insecure: <true|false>

# Optional: a valid kubeConfig file provided using a config map
auth: <string>

See configs.go for how these properties are used.

kubernetes_state_source

Collects metrics on the state of deployments, daemonsets, statefulsets, hpas, jobs, cronjobs and replicasets.

prefix: <string>

kubernetes_cadvisor_source

# We recommend using `kubernetes.cadvisor.` Defaults to empty string.
prefix: <string>

kubernetes_control_plane_source

For more information on control plane metrics, see reference.

# We recommend using `120s`
collection:
    interval: "120s"

prometheus_source

# The URL for a prometheus metrics endpoint. Kubernetes service URLs work across namespaces.
url: <string>

# Optional HTTP configuration
httpConfig:
  [ <ClientConfig> ]

# The source (tag) to set for the metrics collected by this source. Defaults to node name.
source: <string>

telegraf_source

# The list of plugins to be enabled. Empty list defaults to enabling all host plugins.
# Supported host plugins are: mem, net, netstat, linux_sysctl_fs, swap, cpu, disk, diskio, system, kernel, processes
plugins: []

# The configuration specific to a single plugin provided in toml format:
# https://github.com/toml-lang/toml
# parsed using https://github.com/influxdata/toml
conf: |
  [ <Telegraf Plugin Config> ]

See a reference example for details.

systemd_source

# Whether to include systemd task metrics. Defaults to true.
taskMetrics: <true|false>

# Whether to include systemd start time metrics. Defaults to true.
startTimeMetrics: <true|false>

# Whether to include restart metrics. Defaults to false.
restartMetrics: <true|false>

# List of glob patterns. Metrics from matching systemd unit names are reported.
unitAllowList:
- 'docker*'
- 'kubelet*'

# List of glob patterns. Metrics from matching systemd unit names are not reported.
unitDenyList:
- '*mount*'
- 'etc*'

Common properties

Prefix, tags and filters

All sources and sinks support the following common properties:

# An optional dot suffixed prefix for metrics emitted by the sink or source.
prefix: <string>

# A map of key value pairs that are included as point tags on all metrics emitted
# by the sink or source.
tags:
  env: non-production
  region: us-west-2

# Filters to be applied to metrics collected by a source or reported by sinks.
filters:
  # List of glob patterns. Only metrics with names matching the list are reported.
  metricAllowList:
  - 'kube.dns.http.*'
  - 'kube.dns.process.*'

  # List of glob patterns. Metrics with names matching the list are dropped.
  metricDenyList:
  - 'kube.dns.go.*'

  # Map of tag names to list of glob patterns. Only metrics containing tag keys and values matching the list will be reported.
  metricTagAllowList:
    env:
    - 'prod*'
    - 'staging*'

  # Map of tag names to list of glob patterns. Metrics containing these tag keys and values will be dropped.
  metricTagDenyList:
    env:
    - 'test*'

  # List of glob patterns. Tags with matching keys will be included. All other tags will be excluded.
  tagInclude:
  - namespace
  - 'label.app'
  - 'label.component'

  # List of glob patterns. Tags with matching keys will be excluded.
  tagExclude:
  - handler
  - image

Custom collection intervals

All sources support using a custom collection interval:

collection:
  # Duration type specified as [0-9]+(ms|[smhdwy])
  interval: 30s

  # Duration type specified as [0-9]+(ms|[smhdwy])
  timeout: 20s
CA cert configuration

To enable the HTTP proxy with CA cert, you will need to create a Kubernetes secret with your CA cert file:

kubectl create secret generic http-proxy-secret -n wavefront --from-file=tls-root-ca-bundle=<YOUR_CA_CERT_FILE>

Then use this example to deploy proxy.

Note: You will need to change YOUR_CLUSTER, YOUR_API_TOKEN, YOUR_HTTP_PROXY_HOST, and YOUR_HTTP_PROXY_PORT in the above example.