Harbor Analytics

Automatically generate and publish PDF usage reports about your Harbor docker registry.

Easy to run, easy to extend.

Harbor Analytics is a set of tools that - together - form a workflow to generate a usage report of a harbor docker registry.

The usual workflow that is incorporated in Harbor Analytics is:

• Download the latest CSV files that represent the current harbor registry
• Perform analytical tasks on the raw data and convert them to human-readable statistics
• Export the statistics as graphical charts
• Generate a PDF containing the charts and descriptions
• Publish the PDF report on a slack channel

Prerequisites

In order to be able to use Harbor Analytics you will need:

• CSV files generated from the harbor database on either an S3 bucket or locally
• Docker (installed on the machine on which you run the workflow)
• Access to a slack token (if you want to report the metrics)
• A workflow runner (e.g. Jenkins - Jenkinsfile in repo) if you want to run the workflow in a controlled environment

Usage

Prepare Data

First we will need the flat CSV data. Either copy them manually INTO the raw folder or run. Make sure they are named correctly (check analyst/parser.go to see what's expected).

make get-raw-data

You may need to run

make build-raw-data-fetcher

before to create the docker image that will pull your CSVs from S3.

Build Tool

This is optional and only necessary if you have performed any changes in the analyst source code. Recompile the go tool with

make build

This will create a new docker image with the analyst tool in it. You can also compile the tool on your machine directly with

cd analyst && make build

This will not create a docker image and instead just provide you with the executable analyst binary.

Configure

You can configure what kind of analysis the tool should perform through the analyst.yaml config file.

The charts item holds a list of items describing the charts to be generated from the raw data and put into the end report. A charts item is described by specifying a stats method (a method of the registry struct which accepts a StatsMethodParameters parameter and returns a outputgen.BarChartable) and their parameters as sub-items.

Run the analysis

make run

This will create the analytics report PDF and output it to the out folder. Note, however, that this will run the dockerised analyst. If you created a binary on you machine, you need to run

cd analyst && ./analyst

instead.

Publish the Report to slack

make publish

This will post the PDF report on to a slack channel.