Sentry alerted tonight for being slave stopped on the backup server. Investigation turned out that sentry was using GTID mode and the slave wasn't. I fixed the slave, checked it into puppet, and it appears to be catching up.
Turned out MIXED mode was also set. Fixed that and checked it into puppet. Replication caught up and is good to go.
What was the fix? I assume you set the backups to MIXED + GTID....just want to state assumptions.
Yes, that was the fix. I added sentry to the MIXED mode and the GTID exceptions lists in the backups.cnf.erb template.