This has been resolved.
The issue was due to an unforeseen case where processing a previously scheduled campaign, or a campaign scheduled at the exact same time as another larger campaign, would take much longer than expected to process (ie, minutes instead of seconds). While resources were allocated to process the first larger campaign, seperate campaign watchdogs (designed as a failsafe solution) would identify campaign that was scheduled for minutes ago has not yet been processed and would trigger their own campaign start procedure to ensure the send happened. The watchdogs would continue to do this until the first campaign send was completed.
**What we have done to remedy this: **
We have identified the exact cause and have been able to replicate the issue within test environments. Using that we have implemented a fix that introduces a locking process to ensure the watchdogs don't interpret campaigns as 'not sent' when they're in fact processing and erroneously re-triggering the campaign. We have also reviewed and added additional processes and additional automated validation in our workflows to ensure this type of issue is not presented again.
