File attachment issues (staging)

[KatharineV]

Something is wrong with Bulkrax on both production and staging in the Adventist environment. The issues are inconsistent and different between prod and staging environments. I'm going to do the best I can to document the issues. Some issues may be related to Ticket #673.

Importer details:

• CSV: Properly formatted CSV import with related_url files to attach from open AWS S3 bucket. Confirmed the links are active and open--clicking them does prompt file download. @ShanaLMoore checked the CSV and saw no errors in the file itself.
• CSV with zipped files: Properly formatted CSV with zipped files that succeeded on ADL production in March 2024, but failed with weird errors now in staging.

Attempt #1: https://adl.s2.adventistdigitallibrary.org/importers/256?locale=en
Importer ran last week and failed 7 out of 10 works with Errno::ENOENT error message. Works that did import failed to attach files. The items list at the bottom of the page shows a failed file set rather than an attached TIFF.
Shana ran the importer again today (6/17/24) and 7 out of 10 works failed with the same Errno message, and an 8th work failed for TypeError. The new Errno failure was a work that did import in a previous attempt, although without attaching the file.

Attempt #2: https://adl.s2.adventistdigitallibrary.org/importers/257?locale=en
I recreated the CSV after passing the metadata through notepad to strip any invisible characters. I also double-checked the AWS links and re-affirmed that files download.
Importer failed the same 7 out of 10 works with Errno::ENOENT error message. Works that did import failed to attach files. The items list at the bottom of the page shows a failed file set rather than an attached TIFF.

Attempt #3: https://adl.s2.adventistdigitallibrary.org/importers/258?locale=en
I used a CSV and zipped files that had previously fully succeeded in production in March. This June in staging, the Errno::ENOENT error caused half the 10 works to fail. The other half succeeded and did attach files from the zipped folder.

So it seems like the related_url field is having more problems than zipped files dropped into the server, but all importers are having weird errors.