ActiveStorage AWS resource error on assets precompile

[cmaxw]

I'm deploying with Kamal... I'm getting this error when I try to deploy

bundle exec rails assets:precompile
#13 4.658 Error retrieving instance profile credentials: Failed to open TCP connection to 169.254.169.254:80 (Connection refused - connect(2) for "169.254.169.254" port 80)
#13 4.684 bin/rails aborted!
#13 4.697 ArgumentError: missing required option :name (ArgumentError)
#13 4.697
#13 4.697       when nil then raise ArgumentError, "missing required option :name"
#13 4.697                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
#13 4.697 /rails/app/models/meetup.rb:19:in `<class:Meetup>'
#13 4.697 /rails/app/models/meetup.rb:18:in `<main>'
#13 4.697 /rails/app/avo/resources/meetup_replay.rb:4:in `<class:MeetupReplay>'
#13 4.697 /rails/app/avo/resources/meetup_replay.rb:1:in `<main>'
#13 4.697 /rails/config/environment.rb:7:in `<main>'
#13 4.697 Tasks: TOP => assets:precompile => environment
#13 4.697 (See full trace by running task with --trace)
#13 ERROR: process "/bin/sh -c SECRET_KEY_BASE_DUMMY=1 bundle exec rails assets:precompile" did not complete successfully: exit code: 1
------
 > [8/8] RUN SECRET_KEY_BASE_DUMMY=1 bundle exec rails assets:precompile:
4.697
4.697       when nil then raise ArgumentError, "missing required option :name"
4.697                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
4.697 /rails/app/models/meetup.rb:19:in `<class:Meetup>'
4.697 /rails/app/models/meetup.rb:18:in `<main>'
4.697 /rails/app/avo/resources/meetup_replay.rb:4:in `<class:MeetupReplay>'
4.697 /rails/app/avo/resources/meetup_replay.rb:1:in `<main>'
4.697 /rails/config/environment.rb:7:in `<main>'
4.697 Tasks: TOP => assets:precompile => environment
4.697 (See full trace by running task with --trace)
------
Dockerfile:49
--------------------
  47 |
  48 |     # Precompiling assets for production without requiring secret RAILS_MASTER_KEY
  49 | >>> RUN SECRET_KEY_BASE_DUMMY=1 bundle exec rails assets:precompile
  50 |
  51 |     # Entrypoint prepares the database.
--------------------

The issue tracks back to this resource:

class Avo::Resources::MeetupReplay < Avo::BaseResource
  # self.includes = []
  # self.attachments = []
  self.model_class = ::Meetup
  self.index_query = -> { query.past }
  self.default_view_type = :grid
  # self.search = {
  #   query: -> { query.ransack(id_eq: params[:q], m: "or").result(distinct: false) }
  # }

  def fields
    field :id, as: :idlo
    field :title, as: :text
    field :slug, as: :text
    field :description, as: :textarea
    field :starts_at, as: :date_time
    field :video_embed, as: :textarea
    field :participant_url, as: :text
    field :location, as: :text
    field :thumbnail, as: :file
    field :authors, as: :has_many, through: :meetup_authors
    field :membership_level_meetups, as: :has_many
    field :membership_levels, as: :has_many, through: :membership_level_meetups
    field :transcript, as: :trix
    field :notes, as: :trix
  end

  self.grid_view = {
    card: -> do
      {
        cover_url:
          if record.thumbnail.attached?
            main_app.url_for(record.thumbnail.url)
          end,
        title: record.title,
        body: record.description&.truncate(60, separator: /\s/)
      }
    end
  }
end

The error points to the has_one_attached :thumbnail

It works on the meetup resource, but not on the MeetupReplay resource. The only difference between the two files is the model_class and the index_query scope.

It's also working in development, which uses :disk for storage instead of an AWS compatible storage.

[adrianthedev]

Hey @cmaxw please post the model source code as well.

I see that the stack trace points to the model, not the resource.
I would be inclined to say it's not an Avo problem.

[cmaxw]

Everything works fine when I take that resource out. I can even upload new images and put them into cloud storage.

[cmaxw]

So, the credentials work.

[adrianthedev]

Hmmm. ok.
My hunch is that something is triggering the model to connect to AWS and probably the credentials are not loaded/available in that moment.

So you're saying that other resources that have active storage files attached and don'te have the model_class option work just fine.

Let me dig into it a bit more.
Trying to see if model_class makes things weird.

[Paul-Bob]

Everything works fine when I take that resource out. I can even upload new images and put them into cloud storage.

Maybe comment element by element to identify which field or configuration is causing the error (I guess field :thumbnail, as: :file).

I just noticed that your :id field is declared as :idlo, which is a non-existent type. While this might not be the cause of the issue, you should change it to field :id, as: :id.

[Paul-Bob]

Is there any option that configures a bucket name? The error complains about missing :name ("missing required option :name")

[cmaxw]

Correct. I have a Meetup resource that is the same except the index_query and model_class

[adrianthedev]

Got it, Charles!

We're going to be out on vacation next week but we'll try to debug this when we get some spare time.

[adrianthedev]

Actually, please add this bit of code somewhere in an initializer.
Could be avo.rb at the bottom.

module Avo
  module Resources
    class Base
      def self.model_class=(value)
        @model_class = case value
        when Class
          value
        when String, Symbol
          value.to_s
        else
          raise ArgumentError.new "Failed to find a proper model class for #{self}"
        end
      end
    end
  end
end

I'm trying to see if this will make that step of the way pass. Avo might fail somewhere else, but at least we validated that this was causing your error.

[cmaxw]

It deploys now... but I'm getting this error when I load the page:

unable to find policy StringPolicyfor"Meetup" (Avo::NoPolicyError)

[cmaxw]

I get the same when I have the string "::Meetup"

[cmaxw]

If I switch it back to a class like Meetup or ::Meetup it gives the same error there, even with the initializer changed up.

[cmaxw]

I finally gave up and created a new model. I could also have done it with scopes if I upgraded to Advanced... but Replays are different enough from Meetups that I'm fine with what I've got.

[adrianthedev]

Hmmm. This isn't what we intended. Your domain model should not have to suffer because of Avo.

I think I found the "bug". We are trying to constantize the model_class when we set it and that's triggering the error.
We will refactor it to constantize it when it's requested. I'm opening an issue for that.

Thanks for the troubleshooting session.

[adrianthedev]

Fixed by #3207