README
¶
Datadog Cluster Agent - DCA
The Datadog Cluster Agent (referred to as DCA) faithfully collects events and metrics and brings them to Datadog on your behalf so that you can do something useful with your monitoring and performance data.
The purpose of the DCA is to be used alongside of an orchestrator. So far, only Kubernetes is supported. Without the DCA, node agents would have to hit the API Server, which would apply an important pressure on it especially in large clusters.
The DCA has two goals:
- Be the main interface with the API server to collect and forward events.
- Implement the backend for applications requiring a single interface, for instance:
- Keep a map of container and their metadata associated that would otherwise need to be queried by each agent to the API Server.
- Collect the Control Plane health check
The present repository contains the source code of the Datadog Cluster Agent version 6, currently in Alpha.
Getting started
For pre-requisite, refer to the Agent's Getting Started section in the README
To start working on the Cluster Agent, you can build the main branch:
- Clone the repo:
git clone https://github.com/DataDog/datadog-agent.git $GOPATH/src/github.com/DataDog/datadog-agent. - cd into the project folder:
cd $GOPATH/src/github.com/DataDog/datadog-agent. - Install go tools:
dda inv install-tools. - Build the whole project with
dda inv cluster-agent.build
Please refer to the Agent Developer Guide for more details.
Run
To start the agent type cluster-agent start from the bin/datadog-cluster-agent folder, it will take
care of adjusting paths and run the binary in foreground.
You need to provide a valid API key, either through the config file or passing the environment variable like:
DD_API_KEY=12345678990 ./bin/datadog-cluster-agent/datadog-cluster-agent
Features
Once built, you can use the start command and the DCA will also try to connect to the API Server.
If successful, it will forward the events from the API Server to your Datadog app and a health check for each component of the control pane.
Secondly, it will start serving the DD_CLUSTER_AGENT.CMD_PORT if set or 5005 by default with the following endpoints:
- /hostname
- /version
- /api/v1/{check}/checks (available for Kubernetes only in 6.0.0)
- /api/v1/metadata/{host}/{container:[0-9a-z]{64}} (returning the metadata of the said source available in the API Server)
- /flare
- /stop
- /status
Documentation
The general documentation of the project (including instructions on the Beta builds, Agent installation, development, etc) is located under the docs directory of the present repo.
Contributing code
You must sign a CLA before we can accept your contributions. If you submit a PR without having signed it, our bot will prompt you to do so. Once signed you will be set for all contributions going forward.
Pre-requisites for the DCA to interact with the API server.
For the DCA to produce events, service checks and run checks one needs to enable it to perform a few actions. Please find the minimum RBAC here to get the full scope of features. This manifest will create a Service Account, a Cluster Role with a restricted scope and actions detailed below and a Cluster Role Binding as well.
Documentation
¶
Overview ¶
Package clusteragent contains the functionality of the Cluster Agent.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type ServerContext ¶
type ServerContext struct {
ClusterCheckHandler *clusterchecks.Handler
}
ServerContext holds business logic classes required to setup API endpoints
Source Files
Directories
¶
| Path | Synopsis |
|---|---|
|
Package api contains the telemetry of the Cluster Agent API and implements the forwarding of queries from Cluster Agent followers to the leader.
|
Package api contains the telemetry of the Cluster Agent API and implements the forwarding of queries from Cluster Agent followers to the leader. |
|
v1
Package v1 contains the types of the Cluster Agent API (v1).
|
Package v1 contains the types of the Cluster Agent API (v1). |
|
Package autoscaling implements common components for all autoscaling controllers.
|
Package autoscaling implements common components for all autoscaling controllers. |
|
cluster
Package cluster contains the controller for cluster autoscaling.
|
Package cluster contains the controller for cluster autoscaling. |
|
cluster/model
Package model contains the data model for cluster autoscaling.
|
Package model contains the data model for cluster autoscaling. |
|
custommetrics
Package custommetrics implements the part of the Cluster Agent that's responsible for providing custom metrics to the Kubernetes apiserver for Horizontal Pod Autoscalers.
|
Package custommetrics implements the part of the Cluster Agent that's responsible for providing custom metrics to the Kubernetes apiserver for Horizontal Pod Autoscalers. |
|
externalmetrics
Package externalmetrics implements the part of the Cluster Agent that's responsible for providing custom metrics to the Kubernetes apiserver for Horizontal Pod Autoscalers.
|
Package externalmetrics implements the part of the Cluster Agent that's responsible for providing custom metrics to the Kubernetes apiserver for Horizontal Pod Autoscalers. |
|
workload
Package workload implements the DatadogPodAutoscaler controller and related components to implement horizontal and vertical pod autoscaling.
|
Package workload implements the DatadogPodAutoscaler controller and related components to implement horizontal and vertical pod autoscaling. |
|
workload/common
Package common implements shared structs used for workload autoscaling.
|
Package common implements shared structs used for workload autoscaling. |
|
workload/external
Package external implements external workload autoscaling.
|
Package external implements external workload autoscaling. |
|
workload/loadstore
Package loadstore provides a noop implementation for the autoscaling workload check.
|
Package loadstore provides a noop implementation for the autoscaling workload check. |
|
workload/local
Package local implements local workload autoscaling.
|
Package local implements local workload autoscaling. |
|
workload/model
Package model implements data model structures and helpers for workload autoscaling.
|
Package model implements data model structures and helpers for workload autoscaling. |
|
workload/provider
Package provider contains logic to start workload autoscaling controllers and provide recommendations
|
Package provider contains logic to start workload autoscaling controllers and provide recommendations |
|
types
Package types implements the types used by the Cluster checks dispatching functionality.
|
Package types implements the types used by the Cluster checks dispatching functionality. |
|
Package languagedetection implements the language detection patcher.
|
Package languagedetection implements the language detection patcher. |