Monocle

The main idea behind Monocle is to detect anomalies in the way changes are produced in your project on GitHub and Gerrit.

Each team is unique in its way of working: how to do code reviews, how many reviewers, how to do CI, is self merge allowed...

So the philosophy behind Monocle is to let you visualize and explore metrics and data that are relevant to the way you work by navigating and filtering in the web user interface.

Browse changes metrics on the demo instance. Follow us on @change_metrics on Twitter.

For example, your team may want to know:

• what is the ratio of created changes vs merged changes?
• is there a good balance between change creations and change reviews?
• what is the ratio of abandoned changes?
• what are the collaboration patterns between the team members?
• how long does it take to merge a change?
• average delay for the first comment or review?
• long standing changes?
• do we have new contributors?

Here is a graph representing what has been going on during a period of time:

Here is the graph of collaboration patterns:

Here is the graph of the complexity versus time to merge changes:

Components

Monocle supports GitHub Pull Requests and Gerrit Reviews. Monocle provides a set of crawlers designed to fetch Pull Requests and Reviews data from the GitHub or Gerrit APIs and to store changes and changes related events into Elasticsearch. These changes and events are exposed via a JSON API. Furthermore Monocle implements ready to use queries and a React based web UI.

To summarize we have the following components in Monocle:

1. an Elasticsearch data store.
2. an api service.
3. a crawler service.
4. a web UI.

Installation

Monocle is in an early phase of development. Feedback is highly welcome.

The process below describes how to index changes from a GitHub repository, a full GitHub organisation and Gerrit repositories, and then how to start the web UI to browse metrics using docker-compose.

Clone and create the needed directories

$ git clone https://github.com/change-metrics/monocle.git
$ cd monocle
$ ln -s docker-compose.yml.img docker-compose.yml

By default docker-compose will fetch the latest published container images. Indeed, we produce Docker container images for the master version of Monocle. If running master does not git your needs, you could still use the last release by setting the MONOCLE_VERSION to 0.8.1 in the .env file. Please refer to Configuration of the containers.

Create the config.yaml file

The config.yaml file is used by the crawler and api services.

If you want to crawl GitHub public repositories, generate a personal access token on GitHub (w/o any specific rights) at https://github.com/settings/tokens. In case of GitHub private repositories, see the GitHub private repositories section.

Then create the config file etc/config.yaml. Here is an example your could start with. Make sure to replace <github_token> by your personal access token:

---
tenants:
  - index: monocle
    crawler:
      loop_delay: 10
      github_orgs:
        - name: tektoncd
          repository: pipeline
          updated_since: "2020-05-01"
          token: <github_token>
          base_url: https://github.com

To crawl the full tektoncd GitHub organization then remove the repository entry from the file. A more complete example is available in the section Full configuration file example.

Start docker-compose

Start Monocle:

$ docker-compose up -d

Ensure services are running:

$ docker-compose ps
monocle_api_1       uwsgi --uid guest --gid no ...   Up      0.0.0.0:9876->9876/tcp, 0.0.0.0:9877->9877/tcp
monocle_crawler_1   monocle --elastic-conn ela ...   Up
monocle_elastic_1   /usr/local/bin/docker-entr ...   Up      0.0.0.0:9200->9200/tcp, 9300/tcp
monocle_web_1       docker-entrypoint.sh /bin/ ...   Up      0.0.0.0:3000->3000/tcp

You might need to check the crawler logs to ensure the crawler started to fetch changes:

$ docker-compose logs -f crawler

You should be able to access the web UI at http://localhost:3000.

After a change in the configuration file, the api and crawler services need to be restarted:

$ docker-compose restart api
$ docker-compose restart crawler

Troubleshooting

ElasticSearch could need some capabilities to run in container mode. Take a look at the logs to see if it started correctly:

$ docker-compose logs elastic

For example, you could need to increase this system parameter:

$ sudo sysctl -w vm.max_map_count=262144

or make the data directory writable for other:

$ chmod o+w data

You might want to wipe a Monocle index:

docker-compose run --rm --no-deps crawler /usr/local/bin/monocle \
--elastic-conn elastic:9200 dbmanage --index <index-name> --delete-repository ".*"

or delete an index:

docker-compose run --rm --no-deps crawler /usr/local/bin/monocle \
--elastic-conn elastic:9200 dbmanage --index <index-name> --delete-index

Advanced deployment configuration

Configuration of the containers

For a local deployment, default settings are fine.

The following settings are available in the .env file:

• MONOCLE_URL=<host or ip> to configure the URL serving the Web UI (default http://localhost:3000).
• MONOCLE_API_URL=<host or ip> to configure the URL serving the API (default http://localhost:9876).
• MONOCLE_VERSION=<version> to use a specific version. By default it uses latest.
• MONOCLE_TITLE=<title> to change the title of the web application. By default it is Monocle.
• ES_XMS and ES_XMX to change the ElasticSearch JVM HEAP SIZE. By default 512m.
• MONOCLE_API_ADDR=<ip> to change the IP address the API service is listening to (default 0.0.0.0).
• MONOCLE_WEB_ADDR=<ip> to change the IP address the Web service is listening to (default 0.0.0.0).
• MONOCLE_ELASTIC_ADDR=<ip> to change the IP address the ElasticSearch service is listening to (default 0.0.0.0). This is only exposed in the development version of the docker-compose (docker-compose.yml.dev).

GitHub authentication

If you want to protect the access to your indices, you can require a GitHub login to access and the people able to use the indices will be the ones listed in the users section in config.yaml.

Configure the GitHub Oauth authentication to secure private indexes

1. Create an Oauth APP in your GitHub user settings page
2. Add "http://$MONOCLE_HOST:9876/api/0/authorize" in "User authorization callback URL"
3. Save the CLIENT_ID and CLIENT_SECRET into .env as GITHUB_CLIENT_ID=<CLIENT_ID> and GITHUB_CLIENT_SECRET=<CLIENT_SECRET>.

The authentication and authorization support is new and only provides a solution to control access to private indexes. Only login users part of users will be authorized to access the related index.

Note that the GitHub application can be also used as a Oauth APP.

GitHub application

Monocle can interact with a GitHub application to create and use installed application token to query the API.

Once the application is created and Monocle started with application id and private key. If a github_orgs entry's token attribute is missing Monocle will search accross the application installations for an installed application on the related GitHub organization. If any, it will generate an installation token for the matching installation and use it to query the GitHub API.

Create the application on GitHub

1. Register new GitHub App
2. In Repository permissions set Metadata as Read-Only, Pull requests as Read-Only and Contents as Read-Only
3. Click Create the GitHub App
4. Click Generate a private key and download the key
5. Save the App ID

Setup Monocle to use the application

1. Save the private key into etc/app_key.rsa
2. Into the .env file add GITHUB_APP_ID=<APP_ID> and GITHUB_APP_KEY_PATH=/etc/monocle/app_key.rsa

GitHub private repositories

To let Monocle crawl and index privates repositories, either you must use a GitHub Application or you must generate a Personal Access Token with the "repo" scope.

Identity Management

Monocle is able to index changes from multiple code review systems. A contributor might get different identities across code review systems. Thus Monocle provides a configuration section to define aliases for contributors.

Let say a Monocle index is configured to fetch changes from github.com and review.opendev.org (Gerrit) and we would like that John's metrics are merged under the John Doe identity.

tenants:
  - index: example
    idents:
      - ident: John Doe
        aliases:
          - github.com/john-doe
          - review.opendev.org/John Doe/12345
    crawler:
      loop_delay: 300
      github_orgs:
        - name: containers
          updated_since: "2000-01-01"
          base_url: https://github.com
      gerrit_repositories:
        - name: ^openstack/.*
          updated_since: "2000-01-01"
          base_url: https://review.opendev.org

A contributor id on github.com or a GitHub enterprise instance is formated as <domain>/<login>.

A contributor id on a Gerrit instance is formated as <domain>/<Full Name>/<gerrit-user-id>.

Apply idents configuration

Database objects must be updated to reflect the configuration. Once config.yaml is updated, run the following commands:

docker-compose stop crawler
docker-compose run --rm --no-deps crawler /usr/local/bin/monocle --elastic-conn elastic:9200 dbmanage --index <index-name> --config /etc/monocle/config.yaml --update-idents
docker-compose restart api
docker-compose start crawler

Full configuration file example

---
tenants:
  - index: monocle
    crawler:
      loop_delay: 10
      github_orgs:
        - name: tektoncd
          repository: pipeline
          updated_since: "2020-03-15"
          token: <github_token>
          base_url: https://github.com
        - name: spinnaker
          updated_since: "2020-03-15"
          token: <github_token>
          base_url: https://github.com
  - index: zuul
    crawler:
      loop_delay: 600
      gerrit_repositories:
        - name: ^zuul/.*
          updated_since: "2020-03-15"
          base_url: https://review.opendev.org
  - index: openstack
    idents:
      - ident: "Fabien Boucher"
        aliases:
          - "review.opendev.org/Fabien Boucher/6889"
          - "review.rdoproject.org/Fabien Boucher/112"
    crawler:
      loop_delay: 600
      gerrit_repositories:
        - name: "^openstack/.*"
          updated_since: "2021-01-01"
          base_url: https://review.opendev.org/
        - name: "^openstack/.*"
          updated_since: "2021-01-01"
          base_url: https://review.rdoproject.org/r/
          prefix: "rdo/"
  # A private index. Only whitelisted users are authorized to access
  # See "Advanced deployment configuration" section
  - index: monocle-private
    users:
      - <github_login1>
      - <github_login2>
    crawler:
      loop_delay: 10
      github_orgs:
        - name: containers
          repository: libpod
          updated_since: "2020-03-15"
          token: <github_token>
          base_url: https://github.com

Database migration

From version 0.8.X to next stable

Identities are consolidated in the database, to enable multiple code review identities (across code review systems) to be grouped.

1. Run the migration process for each index

docker-compose stop
docker-compose start elastic
# For each indexes
docker-compose run --rm --no-deps crawler /usr/local/bin/monocle --elastic-conn elastic:9200 dbmanage --index <index-name> --run-migrate from-0.8-to-last-stable
docker-compose up -d

From version 0.7.0

A new field self_merged has been added. Previously indexed changes can be updated by running the self-merge migration process.

docker-compose run --rm --no-deps crawler /usr/local/bin/monocle --elastic-conn elastic:9200 dbmanage --index <index-name> --run-migrate self-merge

Using external authentication system

Monocle is supporting the "REMOTE_USER" header, which is mostly used by sign-on solutions. When Web server takes care of authentitaction, it set a "REMOTE_USER" environment variable, which can be used by Monocle. To check that, you are able to run simple curl command:

curl --header "REMOTE_USER: Daniel" -XGET http://localhost:9876/api/0/whoami

Contributing

Follow our contributing guide.