I also believed this might be the suspected cause of such failures, but I was wrong.
The Test Climates
With more than three clients I have engaged in extreme measures to isolate issues concerning attachment inconsistencies and failures to process images as well as PDF documents. All of my tests and debugging activities are specific to the use of the Airtable API and Google Apps Script to perform mass extractions and attachment uploads into Airtable.
In 100% of the cases, we ultimately proved that document sources were not the cause. If the sources are properly configured for accessibility and the file URLs are constructed in a manner to reveal the actual artifacts (as opposed to rendering veneers), the Drive platform provided the climate for a successful extraction.
Rate Limiting
We also ruled out rate limiting and other quota breaches by performing mass high-velocity extractions into other systems such as local drives, network drives, and even Amazon S3 object stores. One test included the processing of 10,000 PDF documents from Google Drive into S3 over a five hour period without a single issue; roughly one TB of content.
It’s Airtable
As such, the failures that these three clients experience (even today) with their Airtable solutions are almost certainly in Airtable itself, not the document source platforms. To be clear, this causal assertion is with the Airtable as it pertains to the API only. There may or may not be other attachment issues with manual uploads or with other script processes in Blocks or automations but I can only speak to my experiences with the API.
The Pattern
If we’re talking about the same problem y’all are discussing here, the error condition is unmistakable - an API call into Airtable creates an attachment and for a few seconds it looks like the upload sticks because the UI indicates a document has arrived in the cell. The icon of the document is greyed out during this brief indication period which typically lasts a few seconds. Then the icon vanishes as if to suggest that Airtable tried to store the file but gave up.
Oddly, this behaviour is roughly the identical experience if you use the API to make an attachment using a URL that cannot be accessed by Airtable, a condition that should fail.
The Frequency of the Pattern
These observed failures are intermittent and from time-to-time there are periods where all attachment attempts via the API fail. And there are periods where all attachment attempts succeed. Largely, though, it seems they fail intermittently and randomly.
We also invested heavily in script processes that perform up to ten retries and capture upload success analytics when attachment attempts fail. The analytics from attachment attempts that are retired show equally random outcomes - sometimes a second retry succeeds; sometimes ten retries fail. And in almost every failure situation, a manual upload works.
I believe this issue is still creating problems from time-to-time and in very random cases and it has forced many of my clients to skirt the issue by avoiding attachments and referencing documents by URL only.
