@thisismissem I don't know a lot about your specific situation, but I thought that if you suspected multiple transactions updating the same rows, you might try the SELECT FOR UPDATE approach.
I think a source might be because we've two or three processes that can update data in transactions, maybe without enough indexes, and these get interleaved.
Fixing this means "locking" the "store the latest data" job on the "push the latest data" job (currently these are on separate cron schedules and may overlap)
Add comment