Queue And Media

Holo-JS media can now use the queue subsystem for queued conversions.

Mark a conversion as queued

Use .queued() on the conversion definition:

import { collection, conversion, defineMediaModel } from '@holo-js/media'

export const Post = defineMediaModel(BasePost, {
  collections: [
    collection('images').disk('public'),
  ],
  conversions: [
    conversion('thumb')
      .performOnCollections('images')
      .width(368)
      .height(232)
      .format('webp')
      .queued(),
  ],
})

What happens at runtime

• non-queued conversions still run inline
• queued conversions dispatch the package-owned media.generate-conversions job
• the queued payload contains only serializable identifiers and conversion names
• worker-side processing regenerates the derived files and updates the media row safely

Driver behavior

With the scaffolded sync driver, queued media conversions run immediately because queue dispatch is inline.

With redis or database, the job is enqueued and processed by a worker:

npx holo queue:work --connection redis --queue media

Why the payload stays small

Media conversion jobs do not serialize model instances or file handles. They queue:

• the media row identifier
• the queued conversion names

The worker reloads the media record and original file from storage when it executes the job.

Recommended setup

• keep quick, cheap conversions inline
• queue heavier image work such as large thumbnails or format conversion chains
• run a dedicated media worker if conversions are slower than the rest of your app jobs

Example dedicated worker:

npx holo queue:work --connection redis --queue media

Continue

• Media
• Workers