Re-initialise meilisearch with records in database

[torinfo]

See also issue #3157.

We have a situation where we are sure there are H5P objects in teh database (of the hub and the content-author) but they are not shown in 'Explore' and 'My Content'. this seems to be caused by the fact that the objects are NOT indexed in meilisearch.

Is there a way to re-initialise the database records with meilisearch? For example through 'php artisan scout:import'? If so, what is the complete command line? And should I first do a 'php artisan scout:flush' first?

[tmkarlsen]

Go to /admin and locate the "Rebuild context index". Try using this and see if that helps as a first step.

[torinfo]

Ah, thank you for this. I'll let you know if that solves the issue.

[torinfo]

Question, how long does that take? I started that process about 1 hour ago, and this installation has about 700 projects, and it is still running (at least the bleu box 'Rebuildin content index' is still visible).

[tmkarlsen]

What do you mean by projects? This could be an issue with the message queue handler are you using and probably something you can look into.

[torinfo]

In you screen shot, it mentions 'Active contents in database' as 74390. In this install it is 689. Which message queue handler are you talink about, and how can I check if there is an issue? I've no idea whicj message queue handler this installation uses.

[tmkarlsen]

I see. Then it was as I thought and an amount of under 700 contents should not take long at all - only a couple of minutes really.

[torinfo]

@tmkarlsen Ah, I suppose you mean Redis. I'll check that.

[tmkarlsen]

Correct.

[torinfo]

We're running the snap version of Redis, and it apparently updated itself to version 8.2.1. Could that cause issues?

[emmachughes]

@torinfo Judging by https://github.com/torinfo/Edlib/blob/master/INSTALL_LXD_CONTAINERS.md, you should have a message handling process (php artisan queue:work) set up as a supervisor service. Can you verify this process is receiving messages from Edlib? (It should report running jobs to STDOUT)

Reindexing is quite memory intensive, and the process might also crash due to out-of-memory. If that's the case, you should try increasing PHP's memory limit.

[torinfo]

Thank you so much. So, running php artisan queue:work updated the meiliserach (reinitialised the index after that).
Question still was why it was not running, and it turned out there was a (temporary) hickup with redis that caused so many restarts of the workers that supervisor gave up.

All fixed now. I'll give a comment in #3157 as well, becasue that can be claused as NOT being a bug as well.

Thanks for your help!