Improve the cleaning process for the failed preload actions

[piotrbak]

Before submitting an issue please check that you’ve completed the following steps:

• Made sure you’re on the latest version
• Used the search feature to ensure that the bug hasn’t been reported before

Describe the bug
It looks like our implementation of actions clearing is not taking into the consideration failed actions in the database.

To Reproduce
Steps to reproduce the behavior:

1. Make sure there are some failed preload actions in the AS
2. Wait a couple of minutes or trigger manually rocket_preload_process_pending event.
3. Check the failed actions in the AS, they're not reduced.

Expected behavior
We should also clean the failed actions, as they can stuck in the logs for a very long time. In my case, it looks like AS is not clearing them at all as this one is in the logs for 7 months.

Additional context
Not sure if we should use the same cleaner or different one, it might be useful to have ability to disable this cleaner for debugging purposes.

It's a part of this issue:
#6052

And database health in general can be improved by implementing also those issues:
#6057
#6058

Backlog Grooming (for WP Media dev team use only)

• Reproduce the problem
• Identify the root cause
• Scope a solution
• Estimate the effort

[Mai-Saad]

Note: the same is happening with RUCSS when executing action_scheduler_run_queue_rucss while having failed RUCSS entries in the actions table => only completed jobs are removed from the table

[piotrbak]

Acceptance Criteria:

1. When rocket_preload_process_pending cron is triggered, we should also clean the failed AS jobs related to WP Rocket Preload and RUCSS
2. Cleaner batch size (default 100) should apply for both, failed + completed
3. When rocket_preload_process_pending cron is triggered, we should not clean the failed AS jobs related to 3rd parties
4. With rocket_action_scheduler_retention_period filter, we should be apply to change the lifetime of items in the AS log

[piotrbak]

@Mai-Saad For RUCSS we have

wp-rocket/inc/Engine/Optimization/RUCSS/Cron/Subscriber.php

Line 120 in 9bdfa2f

public function cron_remove_failed_jobs() {

[CrochetFeve0251]

Currently we are using the Cleaner class to clean our AS actions.
However inside that class we didn't listed failed tasks:

wp-rocket/inc/Engine/Common/Queue/Cleaner.php

Line 62 in 9bdfa2f

$statuses_to_purge = [

If we don't want to keep them then we would just have a add them inside that logic.
If we want to have a different lifespan we would need to reimplement the logic from that for loop with a different cutoff:

wp-rocket/inc/Engine/Common/Queue/Cleaner.php

Line 66 in 9bdfa2f

foreach ( $statuses_to_purge as $status ) {

[CrochetFeve0251]

Reproduce the problem

As we already discussed and reproduced the issue during a meeting last week I consider this as done.

Identify the root cause

The root cause is that the cleaner is not clearing errors cf:
https://github.com/wp-media/wp-rocket/blame/219a0989dc6ec41db29b47ac38f0869da57a3e86/inc/Engine/Common/Queue/Cleaner.php#L62

Scope a solution

The solution from @piotrbak is related to the wrong table (wpr table instead of AS table).
Instead for the similar fashion solution in RUCSS we need to look at the RUCSSQueueRunner on this line:

wp-rocket/inc/Engine/Common/Queue/RUCSSQueueRunner.php

Line 75 in 219a098

$cleaner = new Cleaner( $store, $batch_size, $this->group );

As per discussion with @piotrbak we need to support RUCSS failed action if possible and we do not need a specific filter for failed action retention.

Due tot that the solution will be to add the failed actions to the statuses to clean:
https://github.com/wp-media/wp-rocket/blame/219a0989dc6ec41db29b47ac38f0869da57a3e86/inc/Engine/Common/Queue/Cleaner.php#L62

For that we just have to add \ActionScheduler_Store::STATUS_FAILED, to that array.

Estimate the effort

Effort XS

[Tabrisrp]

Looks good to me 👍🏼