Team Configuration

What is 'Team Configuration' and how it works?

To organize and display metrics by team, Apache DevLake needs to know about the team configuration in an organization, specifically:

1. What are the teams?
2. Who are the users(unified identities)?
3. Which users belong to a team?
4. Which accounts(identities in specific tools) belong to the same user?

Each of the questions above corresponds to a table in DevLake's schema, illustrated below:

1. teams table stores all the teams in the organization.
2. users table stores the organization's roster. An entry in the users table corresponds to a person in the org.
3. team_users table stores which users belong to a team.
4. user_accounts table stores which accounts belong to a user. An account refers to an identiy in a DevOps tool and is automatically created when importing data from that tool. For example, a user may have a GitHub account as well as a Jira account.

Apache DevLake uses a simple heuristic algorithm based on emails and names to automatically map accounts to users and populate the user_accounts table. When Apache DevLake cannot confidently map an account to a user due to insufficient information, it allows DevLake users to manually configure the mapping to ensure accuracy and integrity.

A step-by-step guide

In the following sections, we'll walk through how to configure teams and create the five aforementioned tables (teams, users, team_users, accounts, and user_accounts). The overall workflow is:

1. Create the teams table
2. Create the users and team_users table
3. Populate the accounts table via data collection
4. Run a heuristic algorithm to populate user_accounts table
5. Manually update user_accounts when the algorithm can't catch everything

Note:

1. Please replace /path/to/*.csv with the absolute path of the CSV file you'd like to upload.
2. Please replace 127.0.0.1:4000 with your actual Apache DevLake ConfigUI service IP and port number.

Step 1 - Create the teams table

You can create the teams table by sending a PUT request to /plugins/org/teams.csv with a teams.csv file. To jumpstart the process, you can download a template teams.csv from /plugins/org/teams.csv?fake_data=true. Below are the detailed instructions:

a. Download the template teams.csv file

i.  GET http://127.0.0.1:4000/api/plugins/org/teams.csv?fake_data=true (pasting the URL into your browser will download the template)

ii. If you prefer using curl:
    curl --location --request GET 'http://127.0.0.1:4000/api/plugins/org/teams.csv?fake_data=true'

b. Fill out teams.csv file and upload it to DevLake (If you are using Excel to modify the CSV file, please save it with UTF-8 encoding. See how)

i. Fill out `teams.csv` with your org data. Please don't modify the column headers or the file suffix.

ii. Upload `teams.csv` to DevLake with the following curl command: 
curl --location --request PUT 'http://127.0.0.1:4000/api/plugins/org/teams.csv' --form 'file=@"/path/to/teams.csv"'

iii. The PUT request would populate the `teams` table with data from `teams.csv` file.
You can connect to the database and verify the data in the `teams` table.
See Appendix for how to connect to the database.

Step 2 - Create the users and team_users table

You can create the users and team_users table by sending a single PUT request to /plugins/org/users.csv with a users.csv file. To jumpstart the process, you can download a template users.csv from /plugins/org/users.csv?fake_data=true. Below are the detailed instructions:

a. Download the template users.csv file

i.  GET http://127.0.0.1:4000/api/plugins/org/users.csv?fake_data=true (pasting the URL into your browser will download the template)

ii. If you prefer using curl:
curl --location --request GET 'http://127.0.0.1:4000/api/plugins/org/users.csv?fake_data=true'

b. Fill out users.csv and upload to DevLake (If you are using Excel to modify the CSV file, please save it with UTF-8 encoding. See how)

i.  Fill out `users.csv` with your org data. Please don't modify the column headers or the file suffix

ii. Upload `users.csv` to DevLake with the following curl command:
curl --location --request PUT 'http://127.0.0.1:4000/api/plugins/org/users.csv' --form 'file=@"/path/to/users.csv"'

iii. The PUT request would populate the `users` table along with the `team_users` table with data from `users.csv` file.
You can connect to the database and verify these two tables.

c. If you ever want to update team_users or users table, simply upload the updated users.csv to DevLake again following step b.

Step 3 - Populate the accounts table via data collection

The accounts table is automatically populated when you collect data from data sources like GitHub and Jira through DevLake.

For example, the GitHub plugin would create one entry in the accounts table for each GitHub user involved in your repository. For demo purposes, we'll insert some mock data into the accounts table using SQL:

INSERT INTO `accounts` (`id`, `created_at`, `updated_at`, `_raw_data_params`, `_raw_data_table`, `_raw_data_id`, `_raw_data_remark`, `email`, `full_name`, `user_name`, `avatar_url`, `organization`, `created_date`, `status`)
VALUES
        ('github:GithubAccount:1:1234', '2022-07-12 10:54:09.632', '2022-07-12 10:54:09.632', '{\"ConnectionId\":1,\"Owner\":\"apache\",\"Repo\":\"incubator-devlake\"}', '_raw_github_api_pull_request_reviews', 28, '', 'TyroneKCummings@teleworm.us', '', 'Tyrone K. Cummings', 'https://avatars.githubusercontent.com/u/101256042?u=a6e460fbaffce7514cbd65ac739a985f5158dabc&v=4', '', NULL, 0),
        ('jira:JiraAccount:1:629cdf', '2022-07-12 10:54:09.632', '2022-07-12 10:54:09.632', '{\"ConnectionId\":1,\"BoardId\":\"76\"}', '_raw_jira_api_users', 5, '', 'DorothyRUpdegraff@dayrep.com', '', 'Dorothy R. Updegraff', 'https://avatars.jiraxxxx158dabc&v=4', '', NULL, 0);

Step 4 - Run a heuristic algorithm to populate user_accounts table

Now that we have data in both the users and accounts table, we can tell DevLake to infer the mappings between users and accounts with a simple heuristic algorithm based on names and emails.

a. Send an API request to DevLake to run the mapping algorithm

curl --location --request POST '127.0.0.1:4000/api/pipelines' \
--header 'Content-Type: application/json' \
--data-raw '{
    "name": "test",
    "plan":[
        [
            {
                "plugin": "org",
                "subtasks":["connectUserAccountsExact"]
            }
        ]
    ]
}'

b. After successful execution, you can verify the data in user_accounts in the database.

Step 5 - Manually update user_accounts when the algorithm can't catch everything

It is recommended to examine the generated user_accounts table after running the algorithm. We'll demonstrate how to manually update user_accounts when the mapping is inaccurate/incomplete in this section. To make manual verification easier, DevLake provides an API for users to download user_accounts as a CSV file. Alternatively, you can verify and modify user_accounts all by SQL, see Appendix for more info.

a. GET http://127.0.0.1:4000/api/plugins/org/user_account_mapping.csv(pasting the URL into your browser will download the file). If you prefer using curl:

curl --location --request GET 'http://127.0.0.1:4000/api/plugins/org/user_account_mapping.csv'

b. If you find the mapping inaccurate or incomplete, you can modify the user_account_mapping.csv file and then upload it to DevLake. For example, here we change the UserId of row 'Id=github:GithubAccount:1:1234' in the user_account_mapping.csv file to 2.

c. Save and upload the updated user_account_mapping.csv file with the following curl command (If you are using Excel to modify the CSV file, please save it with UTF-8 encoding. See how):

curl --location --request PUT 'http://127.0.0.1:4000/api/plugins/org/user_account_mapping.csv' --form 'file=@"/path/to/user_account_mapping.csv"'

d. You can verify the data in the user_accounts table has been updated.

Appendix A: how to connect to the database

Here we use MySQL as an example. You can install database management tools like Sequel Ace, DataGrip, MySQLWorkbench, etc.

Or through the command line:

mysql -h <ip> -u <username> -p -P <port>

Appendix B: how to examine user_accounts via SQL

SELECT a.id as account_id, a.email, a.user_name as account_user_name, u.id as user_id, u.name as real_name
FROM accounts a
        join user_accounts ua on a.id = ua.account_id
        join users u on ua.user_id = u.id

Troubleshooting

If you run into any problem, please check the Troubleshooting or create an issue