es-sidecar-service/

About

es-sidecar-service is a service that runs alongside elasticsearch. A single instance is intended to service multiple ES instances, both to provide coordination of es-related activities across an ES cluster (and not per-ES-node) and because this is a low-traffic/low-load service.

It provides common ES functionality via a GRPC interface. At present, this is:

• PurgeTimeSeriesIndicesByAge: delete time-series indices older than the provided age
• PurgeDocumentsFromIndexByAge: deletion of documents older than provided age from a named index
  • TODO: when compliance is updated to implement purge behavior, it will need to implement this call in addition to PurgeTimeSeriesIndicesByAge for compatibility with A1. reference: https://github.com/chef/automate/pull/1264/files

This component also provides monitoring of ES node disk usage, and logs alerts when critical thresholds are reached. This is not yet exposed over GRPC, and no action is taken (beyond logging) when disk usage of any node exceeds the configured threshold.

Future Plans

At time of this writing, this service is expected to offer snapshot functionality, and manage multiple snapshot requests from different domain services such only one is executing at a given time (ES does not support concurrent snapshots).

Dev Tools

The primary dev environment for this project is the A2 habitat studio. cd to the repo root and run hab studio enter. You can then start up a test instance with these commands:

build components/es-sidecar-service/
start_deployment_service
chef-automate dev deploy-some  YOUR_ORIGIN/es-sidecar-service --with-deps

You can run the integration tests in a clean studio with this command:

es_sidecar_service_integration

This project includes CLI commands to assist in development and local testing. They're housed in a single file, cmd/es-sidecar/commands/dev.go:

• ./es-sidecar-service grpc will show the available GRPC calls and allow you to execute them from the shell
• ./es-sidecar-service dev will show the other development helper tools.

These tools assume a locally running ES6 instance, which you can bring up in docker via make start-es6. You can stop and purge es6 testing data with make stop-es6 and make purge-es6 respectively.