InboxValidator & InboxWorker still pending in Horizon

[raphgilles]

Hello,

Since march 2020, everytime I try to follow a Pixelfed account with an account from an other instance (Mastodon or Pixelfed), the InboxValidator & InboxWorker jobs are created in Horizon but still pending. So the following is not validated.

I tried several times to create multiple fresh instances on fresh servers, with fresh domains, following the official doc step by step. Like this one I created yesterday : https://beta.bokeh.social/test

Horizon is runing fine, the queue:work is runing fine, all the other jobs are working fine, the self-diagnosis is ok, and I have nothing in the Laravel log or Nginx log.

It's a very recursive issue on a lot of active instances. You can try to follow all these accounts and it will not work : https://cliq.buzz/team https://anar.chi.st/bugs https://pixel.nogafam.es/admin https://instamatic.mountaintown.video/ajroach https://pixelfett.de/thomas https://instafed.cattube.org/spla https://pixel.misterbanal.net/stacyharper https://imas.photo/barippi https://pixels.helpchangethe.world/SpunkyAlpaca https://pixelfed.earth/tscho ect ect

My configuration : Ubuntu 20.04 Nginx 1.18.0 MariadDB 10.3.25 Php 7.4.3 Supervisor for the processes

[johndevs]

I can confirm this, I have my own instances of Mastodon and Pixelfed and my Mastodon messages all get stuck in pixelfeds Pending Jobs queue for Inbox Worker.

I think this started happening end of last year.

Everything is looking green on the Dashboard and Horizon looks okay.

If I ssh into the Horizon container and manually process the queues with

php artisan queue:work --queue=shared

It does process the queue successfully and the status appears in my feed.

Using: Mastodon 4.1.0 Pixelfed 2023-04-12

[johndevs]

I can confirm this, I have my own instances of Mastodon and Pixelfed and my Mastodon messages all get stuck in pixelfeds Pending Jobs queue for Inbox Worker.

I think this started happening end of last year.

Everything is looking green on the Dashboard and Horizon looks okay.

If I ssh into the Horizon container and manually process the queues with

php artisan queue:work --queue=shared

It does process the queue successfully and the status appears in my feed.

Using: Mastodon 4.1.0 Pixelfed 2023-04-12

Managed to somehow fix it by ssh:ing into the worker container and doing

php artisan instance:actor
php artisan horizon:install
php artisan horizon:publish

And then restarting the container.

Seems a bit flaky this, shouldn't the worker script initialize it properly from the start?

[dansup]

Fixed!