Home

Awesome

CircleCI GitHub tag (latest by date) License Total alerts Language grade: JavaScript

APIUICLI
Docker Pulls APIDocker Pulls UIDocker Pulls CLI oclif NPM Downloads/week
<h1 align="center"> Jira Agile Velocity </h1><br> <p align="center"> Builds various sets of Agile metrics and dashboards by fetching data from Jira REST API </p> <p align="center"> <a href="https://github.com/fgerthoffert/jira-agile-velocity" target="_blank"><img alt="Issues View" title="Issues view" src="./docs/jira-velocity-ui.png" width="640" /></a> </p>

Table of Contents

Introduction

Jira's many, many features are a strength, but can also sometimes make it difficult to consume/visualize metrics. This (opinionated) tool was created to provide different (though simple) views of the data currently hosted within your Jira instance. It borrows concepts and code from ZenCrepes and applies it to Jira specificities.

The tool focuses on two main areas:

Jira-agile-velocity aims at being data-centric, leaving the interpretation to the user. The core concept used to derive metrics is simple: An activity has only two states, it either has to be done (remaining work), or it has been done. By specifying the appropriate JQL queries in the configuration, you can define what is considered done and what is considered remaining.

The tool is broken down in 3 codebases in a monorepo configuration, a UI, an API and a CLI (also called jira-agile-velocity on npm). The CLI's role is to fetch data, compute metrics, and generate a JSON payload, while the UI & API are only there to ease consumption of these metrics in a user-friendly manner.

At the moment, the tool is storing its data in json or ndjson files directly on the filesystem. It is sufficient for the current use case and there is no plan to use a database (i.e. MongoDB) on the short term. Instead, future evolutions will likely move closer to ZenCrepes' implementation, with a common (Github, Jira) indexer and all data served by a search-oriented datastore (i.e. Elasticsearch).

Data-Centric

Although much smaller than ZenCrepes, the Jira Agile Velocity tool shares one its core principle: being data-centric and factual, it doesn't account (⚠️ ON PURPOSE ⚠️) for human interpretation.

An example? the Forecast feature provides a view over the coming weeks, but this view (on purpose) doesn't account for vacations or any other type of events that could impact the team's velocity, and it does not (and will not) provide features to include such events in its forecast. As a DM/PM/Scrum Master/Lead/... (pick one), you know your team and can, therefore, adjust estimates based on your experience.

Versions

v0.4.0

Version 0.4.0 introduces significant changes to the codebase and is not backward compatible with previous versions. Altough the approach is similar, cache data will need to be refetched.

This new version is team-centric and supports multiple work streams per team. A large portion of the logic/calculations (such as building completion per week) was moved from the backend to the frontend. This allow for a reduction of the payload.

Previous versions

Please refer to the readme in the corresponding Git Tags for more details about previous versions.

Features

This tool was built to provide a consolidated view over Jira issues and links back to Jira whenever possible. Most charts and elements are clickable, so you can easily open in Jira all issues closed by a team on a particular week, remaining issues in an initiative, ...

Configuration

Metrics displayed in the tool are entirely based upon JQL queries used to identify completed work and fetch a backlog of upcoming activities.

Sample configuration

- name: Easy
  from: '2021-10-01' # How far back to fetch completion metrics (YYYY-MM-DD)
  excludeDays: ['2021-10-26'] # Do not fetch data for these days ['YYYY-MM-DD', 'YYYY-MM-DD']
  streams:
    - name: Initiatives
      completion:
        jql: sprint = Team_A and status changed to Closed ON (##TRANSITION_DATE##)
        childOf: assignee = Team_A and type = initiative
      forecast:
        jql: assignee = Team_A and type = initiative and status != Closed
        fetchChild: true
        effortPct: 50
    - name: Tech Debt
      completion:
        jql: label = Tech and status changed to Closed ON (##TRANSITION_DATE##)
      forecast:
        jql: label = Tech and (sprint in futureSprints() or sprint in openSprints()) and status != Closed
        fetchChild: false
        emptyPoints: 3
        effortPct: 10
    - name: Unclassified
      completion:
        jql: status changed to Closed ON (##TRANSITION_DATE##)
      forecast:
        jql: (sprint in futureSprints() or sprint in openSprints()) and status != Closed
        fetchChild: false
        emptyPoints: 3
        effortPct: 0

Streams

In the tool, a stream is a type of work carried out by a team, this can be used to identify effort spent on various type of tasks (for example developing new features vs addressing technical debt vs miscenaleous).

Streams are mutually exclusive from top to bottom. In the example above, an issue accounted for in the Initiatives stream will not be counted in Tech Debt and Unclassified even if it is returned by their JQL queries. This ensure an issue is only counted once.

Defining completion

Large features are typically organized in Epics, Streams, Initiatives. Which are usually complex to account for in Jira. For this reason two queries can be provided:

For "simpler" issues (for example individual Defects), there is no need to provide a childOf query.

Defining forecast

Forecast build based on a Stream's backlog. The following parameters are available:

User Interface

In general, streams are color-coded using the stream name via a library called material-color-hash.

Weekly Completion

<p align="center"> <a href="https://github.com/fgerthoffert/jira-agile-velocity" target="_blank"><img alt="Weekly Completion" title="Weekly Completion" src="./docs/jira-team-weekly-completion.png" width="480" /></a> </p>

This chart displays the number of points (or issues count) completed, each week per stream. Clicking on the chart opens the corresponding list of tickets in Jira.

In red, a line shows the weekly velocity calculated over a 4 weeks average.

Weekly Velocity

<p align="center"> <a href="https://github.com/fgerthoffert/jira-agile-velocity" target="_blank"><img alt="Weekly Velocity" title="Weekly Velocity" src="./docs/jira-team-weekly-velocity.png" width="480" /></a> </p>

This charts displays for each of the streams the weekly velocity, calculated over a 4 weeks rolling average.

Distribution

<p align="center"> <a href="https://github.com/fgerthoffert/jira-agile-velocity" target="_blank"><img alt="Weekly Distribution" title="Weekly Distribution" src="./docs/jira-team-weekly-distribution.png" width="480" /></a> </p>

This charts displays for each of the streams the effort distribution (in %), calculated over a 4 weeks rolling average. Clicking on the chart opens the corresponding list of tickets in Jira for the 4 weeks used in the calculation.

Review and Plan

<p align="center"> <a href="https://github.com/fgerthoffert/jira-agile-velocity" target="_blank"><img alt="Review and Plan" title="Review and Plan" src="./docs/jira-team-review-plan.png" width="800" /></a> </p>

This section is primariraly useful to get a sense of the volume of activities in a team's backlog.

Naive forecast

This is a visual representation of the elements present in the review and plan table. Mapping in a calendar view the remaining "known" effort.

Below the naive forecast, a table provides the exact completion status of each initiative, with metrics in both Points and Issue Count, this view can answer the following questions:

The graph view provides a glance of all activities in an Initiative using a tree-like view. Numbers bubble-up for each generation of the tree. Hovering displays the corresponding ticket and clicking on it open it in Jira.

<p align="center"> <img alt="Initiatives Completion Graph" title="Initiatives completion graph" src="./docs/initiatives-graph-view.png" width="640" /> </p>

Jira

The point of the tool is to provide a simple and data-centric view over your project management data, not to build an overly complex and highly customizable system. Trying to build a tool that could work on top of any workflows, without requiring the team to change any of its current Agile practices. But, for various reasons, it has not always been possible and there are occurrences where the tool is going slightly over what I would have hoped in terms of the need for Jira customization.

Map Jira fields

Most of the configuration settings should be fairly straight forward (username, host, ...), what might be slightly more challenging though is to identify the various ID for some of Jira fields (points, epics, ...), which will be different from one installation to the next.

There are many resources online providing instructions to do so, and I'd recommend checking some of them

Some of the fields are pre-installed by Jira Portfolio, but using a custom field of the same type should provide you with a similar feature.

The following fields are used:

NameTypeDescription
Story Pointsnumber (or null)This field is used to calculate the velocity
Original Story Pointsnumber (or null)TThis field is used to calculate velocity in the absence of story points, typically this field is used by the team to provide a high-level estimate when planning an entire feature before actually breaking it down in smaller chunks
Parent Initiativestring (or null)Attach a ticket (child) to an initiative (parent), this string is the ticket key (JIRA-1234)
Parent Epicstring (or null)Attach a ticket (child) to an epic (parent), this string is the ticket key (JIRA-1234)

Team user and group

This is one of the elements having an impact on a team's workflow I would have liked to avoid, but creating a dedicated user per team is probably the best option to address some of the challenges with handling the data.

Imagine all of your teams are sharing projects (which is a very typical scenario), you need a way to identify work completed by a team, versus work completed by another one. You can easily achieve this by creating one user group per team and assigning all of your team members to the corresponding group. In that particular case, your JQL queries can be filtered using assignee in membersOf("YOUR-TEAM-GROUP").

Then you also need to identify remaining work assigned to your team, but in the typical Agile fashion, your tickets are likely only assigned to team members when being picked-up just before starting to work on them. What about the rest of your team's sprint (or team backlog)? You could use the sprint name, but this would force a specific format and require some error-prone string parsing (sprints are only defined by their name, you can't attach a sprint to a team for example).

The only "not-too-intrusive" solution I found is to actually create a dedicated user for the team, attach this user to your team group, and assign it your backlog/sprint tickets (you could give it a funny name, I like to call those users "sprinters", like "Team A sprinter"). Then, once the team members begin implementation they'd assign the ticket to them as they would otherwise.

There needs to be a match between this team user' username (the display name can be different) and the actual team name you define in the tool's configuration. For its internal processing, the tool will always use an ID derived from both the team's name in the configuration file and the team's username in Jira. The ID is generated by stripping all non-alphanumerical characters and converting the resulting string to lowercase.

export const getTeamId = (teamName: string) => {
  return String(teamName)
    .replace(/[^a-z0-9+]+/gi, '')
    .toLowerCase();
};
Configuration fileJira UsernameJira Display nameInternal Team ID
Agile Team Aagile-team-aTeam A Sprinteragileteama
Team Londonteam-londonThe Sprintersteamlondon

Hopefully that makes sense.

Daily Completion

Jira doesn't have a key concept of completion directly attached to an issue. It does have a created date, and updated date, but it doesn't have the notion of closed date or completed date in its data model.

As a result, the tool fetches completion data daily by using the date a transition happened, and this has to be built within your query. For example, the tool will run queries such as assignee in membersOf("agile-team") AND status changed to Done ON ('2019-08-01') to fetch all tickets that were completed on a August 1st 2019.

When configuring jqlCompletion, do the following:

Example of valid queries for the jqlCompletion field:

Authentication

When configuring the tool you can choose between anonymous access (no authentication nor authorization) or using Auth0 as an authentication/authorization provider.

The tool ships with a default Auth0 configuration good enough to get started for development purposes (⚠️do not use those for production ⚠️). It will just validate that users are properly authenticated to Google.

Anonymous access

Anonymous access is done by bypassing the authentication layer entirely. If Auth0 is disabled, all requests to the API will be accepted and the UI will not display the login views. Don't worry though, your source data and configuration settings are still safe, the API is only exposing the consolidated payloads for velocity and initiatives.

Auth0 access (Authenticated and authorized)

Access to the tool's UI and API can be secured using Auth0 as an authentication and authorization provider.

The power of using a provider such as Auth0 (versus directly integrating with Google OAuth2 for example) is that it already provides a robust authorization logic (some might say it's not a good thing), removing the need for building such logic within our tool. Auth0 can then be configured to allow only a pre-defined list of users, an entire domain, ... with a scripting language to define and configure the authorization logic (see the Rules section in your Auth0 dashboard). Pretty powerful and easy to setup.

Create an Auth0 application

You will first need to create a "Single Page Application" in the Applications menu, in the settings of your application:

Create an Auth0 API

Then go and create an API in the configuration panel.

Those settings (AUTH0_DOMAIN, AUTH0_CLIENT_ID) are to be set in your docker-compose.yml file.

Install

Docker and docker-compose

All three components have been dockerized and can be easily spun-up using docker-compose. This is the recommended setup to get everything running quickly.

During the first startup, the system will initialize a configuration file (config.yml) in your device folder, this will need to be updated with the desired settings before re-starting the container. Once updated, you can either re-start the environment docker-compose down; docker-compose up or manually trigger a data refresh in the cli container.

docker-compose.yml

You can use the docker-compose file below to spin-up the environment, the only required action is to create a directory on the Docker host's filesystem to host the configuration file and the application cache.

version: '3.7'

volumes:
  data-volume:
    driver: local
    driver_opts:
      type: none
      device: /tmp/jav-data
      o: bind

services:
  jav-cli:
    image: fgerthoffert/jav-cli:latest
    environment:
      - 'CONFIG_DIR=/root/jav-data'
    volumes:
      - data-volume:/root/jav-data

  jav-api:
    image: fgerthoffert/jav-api:latest
    ports:
      - '5001:3001'
    environment:
      - 'CONFIG_PATH=/root/jav-data'
      - 'AUTH0_DISABLED=true'
      - 'AUTH0_DOMAIN=agile-jav.auth0.com'
      - 'AUTH0_AUDIENCE=http://localhost:5001'
    volumes:
      - data-volume:/root/jav-data

  jav-ui:
    image: fgerthoffert/jav-ui:latest
    ports:
      - '5000:80'
    environment:
      - 'API_URL=http://127.0.0.1:5001'
      - 'AUTH0_DISABLED=true'
      - 'AUTH0_DOMAIN=agile-jav.auth0.com'
      - 'AUTH0_CLIENT_ID=sGJDsqzc0VhKgoutaoyLWjyWK9kAwE2f'
      - 'AUTH0_AUDIENCE=http://localhost:5001'

Once the directory is created, you can start the environment:

> mkdir /tmp/jav-data # Replace with the desired directory to store cache and config on host
> docker-compose pull # Fetches the latest version of the containers
> docker-compose up -d # Run the containers in deamon mode (you can remove -d to see log messages)

Specific Windows configuration

On Windows, the configuration file below has to be used and two additional steps are needed.

version: '3.7'

services:
  jav-cli:
    image: fgerthoffert/jav-cli:latest
    environment:
      - 'CONFIG_DIR=/root/jav-data'
    volumes:
      - type: bind
        source: c:/docker-tmp
        target: /root/jav-data

  jav-api:
    image: fgerthoffert/jav-api:latest
    ports:
      - '5001:3001'
    environment:
      - 'CONFIG_PATH=/root/jav-data'
      - 'AUTH0_DISABLED=true'
      - 'AUTH0_DOMAIN=agile-jav.auth0.com'
      - 'AUTH0_AUDIENCE=http://localhost:5001'
    volumes:
      - type: bind
        source: c:/docker-tmp
        target: /root/jav-data

  jav-ui:
    image: fgerthoffert/jav-ui:latest
    ports:
      - '5000:80'
    environment:
      - 'API_URL=http://127.0.0.1:5001'
      - 'AUTH0_DISABLED=true'
      - 'AUTH0_DOMAIN=agile-jav.auth0.com'
      - 'AUTH0_CLIENT_ID=sGJDsqzc0VhKgoutaoyLWjyWK9kAwE2f'
      - 'AUTH0_AUDIENCE=http://localhost:5001'

Note: The cli container is not configured with a cron to refresh data, you'd need to configure this on your host environment (see configuration update below)

Configuration update

To refresh or manually load data, you can run a command directly towards the CLI docker containing using the following command:

> docker exec -it jira-agile-velocity_jav-cli_1 /usr/share/jav/refresh.sh

Replace the container name (jira-agile-velocity_jav-cli_1) with the actual name for the CLI container obtained by executing docker ps.

Access the UI

In its current Docker configuration, the UI is accessible through http://127.0.0.1:5000 (or the port configured for jav-ui). It is strongly recommended to configure a reverse proxy (with nginx for example) serving those resources (UI and API) over HTTPS. But this is considered out of scope of this tool.

Development environment setup

If you are familiar with npm, running the application in development mode should be very straight forward.

> git clone https://github.com/Fgerthoffert/jira-agile-velocity.git
> cd jira-agile-velocity
> cd api
> npm install
> cd ..
> cd ui
> npm install
> cd cli
> npm install
> cd ..

You can now run each application in their own terminal.

CLI

> cd cli
> ./bin/run velocity # Update the velocity metrics
> ./bin/run roadmap # Update the velocity roadmap

API

You should be passing environment variable when starting the API (see example below)

> cd api
> AUTH0_DOMAIN=agile-jav.auth0.com  AUTH0_AUDIENCE=http://localhost:3001 npm run start:dev

The following variables are used:

UI

When developing with the UI, you can update environment variables by editing the .env file in the UI folder

> cd api
> npm run dev

Build workflow

The project's repository is linked to circle-ci for ci/cd.

On Commit

A set of basic checks are executed on every commit, these include linting and, for the CLI app, running a set of minimal commands (used to verify that the app is executable). At a later stage, unit tests might be added (the frameworks are configured to support tests, it's just a matter of spending the time to write them).

On Tag creation

A set of more interesting actions happen when a new tag is created: