An agent designed to retreive metrics from Kubernetes clusters. Runs as a deployment, forwards provided metrics to Circonus.
- Prerequisites
- Installation
- Options - command line parameters
- Dynamic collection - endpoints, services, pods, and nodes
For full functionality a recent version of kube-state-metrics should be installed in the cluster - see kube-state-metrics deployment instructions for more information.
The agent will look for a service named kube-dns
or coredns
in the
kube-system
namespace.
It will check for two annotations on the service that matches:
prometheus.io/scrape
and prometheus.io/port
.
If scrape
is true
, the agent will collect metrics from the endpoints listed
for the target port.
NOTE: If the annotations
prometheus.io/port
andprometheus.io/scrape
are not specified on the DNS service in thekube-system
namespace,enable-dns-metrics
anddns-metrics-port
must be configured via flags, environment variables, or config file in order to receive DNS metrics.
- Clone repo
- In
deploy/default/configuration.yaml
set the following required attributes:- Circonus API Token -
circonus-api-key
- Kubernetes Cluster Name -
kubernetes-name
- short, unique string w/o spaces - Circonus Alert Email -
default-alerts.json
->contact.email
- email address for default alerts
- Circonus API Token -
- Apply
kubectl apply -f deploy/default/
- Clone repo
- Verify
deploy/custom/authrbac.yaml
, alter any applicable settings for cluster security - Change any applicable settings in
deploy/custom/configuration.yaml
, minimum required:- Circonus API Token
- Check Target (optional, Kubernetes cluster name will be used if not supplied) - so agent can find check on restart (short, unique string w/o spaces - normally this is a FQDN)
- Kubernetes Cluster Name - used for check title when creating a check
- Circonus Alert Email - email address for default alerts
- Change any applicable settings in
deploy/custom/deployment.yaml
- Apply
kubectl apply -f deploy/custom/
NOTE: only use this deployment as requested by Circonus
- Clone repo
- In
deploy/observation/configuration.yaml
set the following required attributes:- Circonus API Token -
circonus-api-key
- Kubernetes Cluster Name -
kubernetes-name
- short, unique string w/o spaces
- Circonus API Token -
- Apply
kubectl apply -f deploy/observation/
- Clone repo
- Make updates to files in
contrib/helm
to customize settings - Install
helm install contrib/helm \
--name=<Helm release name> \
--set=circonus_api_key=<some valid Circonus API key> \
--set=kubernetes_name=<kubernetes-name> \
--set=circonus_check_target=<circonus-check-target> \
--wait
Usage:
circonus-kubernetes-agent [flags]
Flags:
--api-app string [ENV: CKA_CIRCONUS_API_APP] Circonus API Token App name (default "circonus-kubernetes-agent")
--api-cafile string [ENV: CKA_CIRCONUS_API_CAFILE] Circonus API CA file
--api-debug [ENV: CKA_CIRCONUS_API_DEBUG] Debug Circonus API calls
--api-key string [ENV: CKA_CIRCONUS_API_KEY] Circonus API Token Key
--api-key-file string [ENV: CKA_CIRCONUS_API_KEY_FILE] Circonus API Token Key file
--api-url string [ENV: CKA_CIRCONUS_API_URL] Circonus API URL (default "https://api.circonus.com/v2/")
--check-broker-ca-file string [ENV: CKA_CIRCONUS_CHECK_BROKER_CAFILE] Circonus Check Broker CA file
--check-broker-cid string [ENV: CKA_CIRCONUS_CHECK_BROKER_CID] Circonus Check Broker CID to use when creating a check (default "/broker/35")
--check-bundle-cid string [ENV: CKA_CIRCONUS_CHECK_BUNDLE_CID] Circonus Check Bundle CID
--check-create [ENV: CKA_CIRCONUS_CHECK_CREATE] Circonus Create Check if needed (default true)
--check-tags string [ENV: CKA_CIRCONUS_CHECK_TAGS] Circonus Check Tags
--check-target string [ENV: CKA_CIRCONUS_CHECK_TARGET] Circonus Check Target host
--check-title string [ENV: CKA_CIRCONUS_CHECK_TITLE] Circonus check display name (default: <--k8s-name> /circonus-kubernetes-agent)
-c, --config string config file (default: /etc/circonus-kubernetes-agent.yaml|.json|.toml)
--custom-rules-file string [ENV: CKA_CIRCONUS_CUSTOM_RULES_FILE] Circonus custom rules configuration file (default "/ck8sa/custom-rules.json")
-d, --debug [ENV: CKA_DEBUG] Enable debug messages
--default-alerts-file string [ENV: CKA_CIRCONUS_DEFAULT_ALERTS_FILE] Circonus default alerts configuration file (default "/ck8sa/default-alerts.json")
--default-streamtags string [ENV: CKA_CIRCONUS_DEFAULT_STREAMTAGS] Circonus default streamtags for all metrics
-h, --help help for circonus-kubernetes-agent
--k8s-api-cafile string [ENV: CKA_K8S_API_CAFILE] Kubernetes API CA File (default "/var/run/secrets/kubernetes.io/serviceaccount/ca.crt")
--k8s-api-timelimit string [ENV: CKA_K8S_API_TIMELIMIT] Kubernetes API request timelimit (default "10s")
--k8s-api-url string [ENV: CKA_K8S_API_URL] Kubernetes API URL (default "https://kubernetes.default.svc")
--k8s-bearer-token string [ENV: CKA_K8S_BEARER_TOKEN] Kubernetes Bearer Token
--k8s-bearer-token-file string [ENV: CKA_K8S_BEARER_TOKEN_FILE] Kubernetes Bearer Token File (default "/var/run/secrets/kubernetes.io/serviceaccount/token")
--k8s-dynamic-collector-file string [ENV: CKA_K8S_DYNAMIC_COLLECTOR_FILE] Kubernetes dynamic collectors configuration file (default "/ck8sa/dynamic-collectors.yaml")
--k8s-enable-api-server [ENV: CKA_K8S_ENABLE_API_SERVER] Kubernetes enable collection from api-server (default true)
--k8s-enable-cadvisor-metrics [ENV: CKA_K8S_ENABLE_CADVISOR_METRICS] Kubernetes enable collection of kubelet cadvisor metrics
--k8s-enable-events [ENV: CKA_K8S_ENABLE_EVENTS] Kubernetes enable collection of events (default true)
--k8s-enable-dns-metrics [ENV: CKA_K8S_ENABLE_DNS_METRICS] Kubernetes enable collection of kube-dns/CoreDNS metrics (default true)
--k8s-enable-kube-state-metrics [ENV: CKA_K8S_ENABLE_KUBE_STATE_METRICS] Kubernetes enable collection from kube-state-metrics (default true)
--k8s-enable-metrics-server [ENV: CKA_K8S_ENABLE_METRICS_SERVER] Kubernetes enable collection from metrics-server
--k8s-enable-node-metrics [ENV: CKA_K8S_ENABLE_NODE_METRICS] Kubernetes include metrics for individual nodes (default true)
--k8s-enable-node-stats [ENV: CKA_K8S_ENABLE_NODE_STATS] Kubernetes include summary stats for individual nodes (and pods) (default true)
--k8s-enable-nodes [ENV: CKA_K8S_ENABLE_NODES] Kubernetes include metrics for individual nodes (default true)
--k8s-include-containers [ENV: CKA_K8S_INCLUDE_CONTAINERS] Kubernetes include metrics for individual containers
--k8s-include-pods [ENV: CKA_K8S_INCLUDE_PODS] Kubernetes include metrics for individual pods (default true)
--k8s-interval string [ENV: CKA_K8S_INTERVAL] Kubernetes Cluster collection interval (default "1m")
--k8s-ksm-field-selector-query string [ENV: CKA_K8S_KSM_FIELD_SELECTOR_QUERY] Kube-state-metrics fieldSelector query for finding the correct KSM installation (default "metadata.name=kube-state-metrics")
--k8s-ksm-metrics-port-name string [ENV: CKA_K8S_KSM_METRICS_PORT_NAME] Kube-state-metrics metrics port name (default "http-metrics")
--k8s-ksm-request-mode string [ENV: CKA_K8S_KSM_REQUEST_MODE] Kube-state-metrics request mode, proxy or direct (default "direct")
--k8s-ksm-telemetry-port-name string [ENV: CKA_K8S_KSM_TELEMETRY_PORT_NAME] Kube-state-metrics telemetry port name (default "telemetry")
--k8s-dns-metrics-port string [ENV: CKA_K8S_DNS_METRICS_PORT] kube-dns/CoreDNS metrics port if annotations not on service definition (default "10054")
--k8s-name string [ENV: CKA_K8S_NAME] Kubernetes Cluster Name (used in check title)
--k8s-node-selector string [ENV: CKA_K8S_NODE_SELECTOR] Kubernetes key:value node label selector expression
--k8s-pod-label-key string [ENV: CKA_K8S_POD_LABEL_KEY] Include pods with label
--k8s-pod-label-val string [ENV: CKA_K8S_POD_LABEL_VAL] Include pods with pod label and matching value
--k8s-pool-size uint [ENV: CKA_K8S_POOL_SIZE] Kubernetes node collector pool size (default 4)
--log-level string [ENV: CKA_LOG_LEVEL] Log level [(panic|fatal|error|warn|info|debug|disabled)] (default "info")
--log-pretty Output formatted/colored log lines [ignored on windows]
--metric-filters-file string [ENV: CKA_CIRCONUS_METRIC_FILTERS_FILE] Circonus Check metric filters configuration file (default "/ck8sa/metric-filters.json")
--show-config string Show config (json|toml|yaml) and exit
--trace-submits string Trace metrics submitted to Circonus to passed directory (one file per submission)
-V, --version Show version and exit
Dynamic collection simplifies gathering desirable metrics exposed in Prometheus format from endpoints, services, pods, and nodes.
- Use Custom deployment in order to configure dynamic collectors
- Add metric filters to the configuration for the desired metrics from dynamic collectors.
The configuraiton is defined in the custom deployment.
collectors:
- name: "" # required
disable: false # disable this collector
type: "" # required - endpoints, nodes, pods, services
selectors: # defaults to all of the type
label: "" # labelSelector expression
field: "" # fieldSelector expression
control: # control whether to collect metrics from specific item
annotation: "" # use the value of an annotation
label: "" # use the value of a label
value: "" # use a static value ("1", "t", "T", "TRUE", "true", "True", "0", "f", "F", "FALSE", "false", "False")
metric_port: # define port to use in metric request
annotation: "" # use the value of an annotation
label: "" # use the value of a label
value: "" # use a static value
metric_path: # define path to use in metric request, default `/metrics`
annotation: "" # use the value of an annotation
label: "" # use the value of a label
value: "" # use a static value
schema: # scema to use in metric request, default "http"
annotation: "" # use the value of an annotation
label: "" # use the value of a label
value: "" # use a static value ("http" or "https")
rollup: # rollup metrics, default "false"
annotation: "" # use the value of an annotation
label: "" # use the value of a label
value: "" # use a static value ("1", "t", "T", "TRUE", "true", "True", "0", "f", "F", "FALSE", "false", "False")
tags: "" # comma separated list of static tags to add
label_tags: "" # comma separated list of labels on the item to add as tags
option | required | description | default |
---|---|---|---|
name | yes | name of this collector | n/a |
disable | no | disable a collector, but keep the configuration | false |
type | yes | type of the collector (endpoints , nodes , pods , services ) |
n/a |
selectors | define what items of the type to collect | ||
selectors.label | no | a labelSelector for the type | all for type |
selectors.field | no | a fieldSelector for the type | all for type |
control | controls collection of metrics (string parsed as boolean) | ||
control.annotation | no | annotation to use e.g. "monitor" |
|
control.label | no | label to use e.g. "metricsCollect" |
|
control.value | no | static value (1 , t , T , TRUE , true , True , 0 , f , F , FALSE , false , False ) |
"true" |
metric_port | defines the port to use for the metric request | ||
metric_port.annotation | no | annotation to use | |
metric_port.label | no | label to use | |
metric_port.value | no | static port for all instances (e.g. "8080" ) |
|
metric_path | defines the path to use for the metric request | ||
metric_path.annotation | no | annotation to use | |
metric_path.label | no | label to use | |
metric_path.value | no | static path to use for all instances | "/metrics" |
schema | no | HTTP request schema http or https |
|
schema.annotation | no | annotation to use | |
schema.label | no | label to use | |
schema.value | no | static schema for all instances | "http" |
rollup | no | control rolling up metrics (string parsed as boolean) | |
rollup.annotation | no | annotation to use | |
rollup.label | no | label to use | |
rollup.value | no | static value (1 , t , T , TRUE , true , True , 0 , f , F , FALSE , false , False ) |
"false" |
tags | no | comma separated list of static tags to add e.g. "app:myapp,foo:bar" |
|
label_tags | no | comma separated list of labels to use as tags e.g. "environment,location" or * to make all labels tags |
From endpoints
with the label k8s-app=kube-dns
collect metrics from port 9153
using the default path of /metrics
.
collectors:
- name: "kube-dns"
type: "endpoints"
selectors:
label: "k8s-app=kube-dns"
metric_port:
value: "9153"
From all pods
with a label appName
set to myapp
, collect if the pod has an annotation metricsCollect
with a value of true
, use the values of the metricsPort
and metricsPath
annotations when constructing the metric request URL. Add the static tag env:prod
and use the item labels appName
and location
as tags.
collectors:
- name: "myapp"
type: "pods"
selectors:
label: "appName=myapp"
control:
annotation: "example.io/metricsCollect"
metric_port:
annotation: "example.io/metricsPort"
metric_path:
annotation: "example.io/metricsPath"
tags: "env:prod"
label_tags: "appName,location"