Refactor #31300
open
the user shouldn't have to specify the file size when uploading chunks
Description
Currently, the workflow to upload a file to a repository is the following:
1. call content_uploads/create with file size and repo_id
2. call content_uploads/update in a loop with the content_upload_id, the actual chunk content, the offset of the chunk and the file size
3. call repositories/import_uploads
4. call content_uploads/destroy
I find the requirement to pass the file size in every chunk upload confusing. I've provided it already when I created the upload in step1, so I'd expect Katello to just reuse that value (it won't change anyways).
Updated by 
Updated by 
Updated by 
Updated by
Updated by Justin Sherrill about 3 years ago
from samir:
The upload chunk is an update call on the upload api in pulp3 with upload_href which takes a content_range(offset, offset + actual_chunk_size - 1, size) param as required value which needs to know the size of the file. I couldn't think of any workarounds to this other than querying the pulp api for size on upload or keeping some sort of db in katello.
https://docs.pulpproject.org/pulpcore/restapi.html#operation/uploads_update
Updated by Evgeni Golov about 3 years ago
it does need it, sure, but can't Katello calculate it? the same way it happens if the upload happens via the UI which does not have a "size" field ;)