Add error handling and basic tracking to abl #13
Conversation
The idea is that we will use partial_data to determine if there was an error. If there was an error, we can cache less greedily.
|
The ABL keeps two versions when someone adds a new version: the version that is currently on REX (if there is one), and the new version they add. Edit: It already does this. |
lib/openstax/content/abl.rb
Outdated
| hash[page.uuid] ||= [] | ||
| hash[page.uuid] << { book: book.slug, page: page.slug } | ||
| end | ||
| rescue StandardError => exception |
There was a problem hiding this comment.
Hmm the issue is that this rescue will prevent the rescue in https://github.com/openstax/content-ruby/blob/main/lib/openstax/content/abl.rb#L57 from triggering, so we'll lose the behavior of trying old archive versions.
Maybe this logic has to replace the raise exception in https://github.com/openstax/content-ruby/blob/main/lib/openstax/content/abl.rb#L58 and https://github.com/openstax/content-ruby/blob/main/lib/openstax/content/abl.rb#L66 instead?
There was a problem hiding this comment.
Oh, you're right. I think I got in too big of a hurry on this one. I will fix it.
There was a problem hiding this comment.
@Dantemss I moved the error handling into each_book_with_previous_archive_version_fallback and changed it so that it tries with the newest archive version each time, and walks backwards up to max_attempts for each book.
Before it would reuse the same previous_archive for subsequent books, and only fail up to max_attempts for all books.
…eful degradation The changes fix a critical bug and significantly improve error handling. Key improvements: 1. Each book now gets its own previous_version and previous_archive instead of reusing the first failing book's archive for all subsequent failures 2. Replace complex batch retry pattern with straightforward per-book retry loop 3. New allow_partial_data parameter allows processing to continue when books fail, logging warnings instead of raising exceptions 4. Remove duplicate try/catch from slugs_by_page_uuid, consolidate in each_book_with_previous_archive_version_fallback 5. Add proper test setup with mocked books to verify graceful degradation and partial data handling
https://openstax.atlassian.net/browse/CORE-1521
The idea is that we will use
partial_datato determine if there was an error. If there was an error, we can cache less greedily.I tried to do this in a way that wouldn't break things that do not care about partial data. It's maybe a little hacky...
Next I plan to update
content.rbin exercises so that it sets the cache expiration to 30 minutes ifpartial_dataistrue.