Add indexes to good_jobs.finished_at and have GoodJob.cleanup_preserved_jobs delete all executions for a given job

[bensheldon]

Closes #476

[6temes]

Hi @bensheldon

I just realized that, after running the latest migrations, I have two indices:

    t.index ["active_job_id", "created_at"], name: "index_good_jobs_on_active_job_id_and_created_at"
    t.index ["active_job_id"], name: "index_good_jobs_on_active_job_id"

I may be wrong, but wouldn't index_good_jobs_on_active_job_id be redundant? If I understand well, index_good_jobs_on_active_job_id_and_created_at is already indexing the active_job_id and, having both indices would only slow down writes.

[bensheldon]

@6temes you're right, those are redundant, and the singular/non-composite index could be dropped.

[nsilva-ta]

Hi @bensheldon

Why do we have an array with only one symbol for the column name?

    add_index :good_jobs, [:finished_at], where: "retried_good_job_id IS NULL AND finished_at IS NOT NULL", name: :index_good_jobs_jobs_on_finished_at

[reczy]

Hi @nsilva-ta,

Both ways of writing in the column_name are acceptable (symbol or array of symbols) - see here.

In fact, if you look at your schema after the migration, you'll see that they're all represented in array format:

    t.index ["active_job_id", "created_at"], name: "index_good_jobs_on_active_job_id_and_created_at"
    t.index ["active_job_id"], name: "index_good_jobs_on_active_job_id"
    t.index ["concurrency_key"], name: "index_good_jobs_on_concurrency_key_when_unfinished", where: "(finished_at IS NULL)"
    t.index ["cron_key", "created_at"], name: "index_good_jobs_on_cron_key_and_created_at"
    t.index ["cron_key", "cron_at"], name: "index_good_jobs_on_cron_key_and_cron_at", unique: true
    t.index ["finished_at"], name: "index_good_jobs_jobs_on_finished_at", where: "((retried_good_job_id IS NULL) AND (finished_at IS NOT NULL))"
    t.index ["queue_name", "scheduled_at"], name: "index_good_jobs_on_queue_name_and_scheduled_at", where: "(finished_at IS NULL)"
    t.index ["scheduled_at"], name: "index_good_jobs_on_scheduled_at", where: "(finished_at IS NULL)"