Open cluster management - Search collector

The search-collector is part of the search component in Open Cluster Management. The search feature spec has an overview of all the search components.

This process targets any kubernetes cluster to collect data about its configuration resources. Then this data is sent to the search-indexer, where it is inserted into a relational database.

Data model

See the Data model documentation for more information.

Features

Search-collector provides a similar solution as Kubernetes controllers and consists of four main components.

• Informer: queries for resources and watches for updates
• Transformer: extracts data from resources to include in the search index and discovers relationships to other resources
• Sender: syncs state and sends changes to the search-indexer
• Reconciler: merges changes into search-collector internal state

Usage and configuration

Running Locally

Pre-requisite: Go v1.18

1. Fetch Dependencies: make deps

   TIP 1: You may need to install mercurial. brew install mercurial

   TIP 2: You may need to configure git to use SSH. Use the following command:

   git config --global --add url."git@github.com:".insteadOf "https://github.com/"

2. Log into your development cluster with oc login ....

   Alternative: set the KUBECONFIG environment variable to some other kubernetes config file.

3. Run the program with make run or go run main.go

Running on a cluster

• Pull image from Quay.io and deploy to your cluster.
• To test code changes, create a PR in this repo, the build process pushes the PR image to Quiay.io. From there you can pull it into your cluster.

Environment Variables

Control the behavior of this service with these environment variables. For development these could be set on ./config.json

Name	Required	Default Value	Description
AGGREGATOR_URL	yes	https://localhost:3010	Deprecated. Use host + port instead.
AGGREGATOR_HOST	yes	https://localhost	Location of the aggregator service.
AGGREGATOR_PORT	yes	3010
CLUSTER_NAME	yes	local-cluster	Name of cluster where this collector is running.
HEARTBEAT_MS	no	300000 // 5 min	Interval(ms) to send empty payload to ensure connection
MAX_BACKOFF_MS	no	600000 // 10 min	Maximum backoff in ms to wait after send error
REDISCOVER_RATE_MS	no	120000 // 2 min	Interval(ms) to poll for changes to CRDs
REPORT_RATE_MS	no	5000 // 5 seconds	Interval(ms) to queue changes before sending to the aggregator
RUNTIME_MODE	no	production	Running mode (development or production)

Other Configuration Options

• Environment variables can also be set in the ./config.json for development. If both provide a value for a specific property, the environment variable overrides the file. You can define your own config.json file and pass it to the application with the following command: -c <config_file>
• The application can take any flags for glog, which passes them straight into glog. The glog flag --logtostderr is set to true by default.

Dev Preview (Search Configurable Collection)

Configurable collection is now fully supported. This topic has moved here.

Contribution

When you make contributions to this project, fork the project then merge your changes by creating a pull request from your personal fork to the main project:

1. Fork the search-collector repository.
2. Clone your fork.
3. Set upstream origin with the following command: git remote add upstream git@github.com:stolostron/search-collector.git
4. Make your edits. Test locally with make test before creating a pull request to ensure smooth merging 😄
5. Push your new commits to your personal fork with the following command: git push origin
6. Create a pull request from your personal fork again the upstream search-collector main branch

Rebuild: 2022-09-16