Load Generator: SQL and NoSQL queries

[dbazhenov]

There are currently 4 groups of SQL/NoSQL queries running for the load. The switches on the control panel allow you to enable and disable each group for each database.

You can see these queries in the files:
/internal/load/mysql.go
/internal/load/postgres.go
/internal/load/mongodb.go

Help is needed to improve these queries so that they are more varied and their complexity grows with each switch.

The following logic is now in place:

1. Switch 1: Get an random row from the Repositories table (200 rows). Write this row to the test table. Delete a row from the test table.
2. Switch 2: Get an random row from the Pulls table (40000 rows). Write this row to the test table. Delete a row from the test table.
3. Switch 3: SELECT repo FROM (SELECT DISTINCT repo FROM pulls) AS uniq_repos ORDER BY RAND() LIMIT 1
4. Switch 4: SELECT data FROM pulls WHERE STR_TO_DATE(JSON_UNQUOTE(JSON_EXTRACT(data, '$.created_at')), '%Y-%m-%dT%H:%i:%sZ') >= NOW() - INTERVAL 3 MONTH LIMIT 10;

I'd be happy to hear your suggestions.

Dataset currently includes JSON documents with repository and pool-request data. JSONs have many fields. I can publish a list if needed, but you can see them in the GitHub API. See the JSON example in the Example response section of the page

• Repositories https://docs.github.com/en/rest/repos/repos?apiVersion=2022-11-28#list-organization-repositories
• Pulls https://docs.github.com/en/rest/pulls/pulls?apiVersion=2022-11-28#list-pull-requests

It is possible to change the table layout, for example to add additional columns. Or additional tables, for example with comments on pull-requests.

[dbazhenov]

Related Issue
#28