...
Schedule 10 minutes of downtime
Put the following in a bash script:
heroku maintenance:on --app scihist-digicoll-production heroku pg:maintenance:run --force --app scihist-digicoll-production heroku pg:wait --app scihist-digicoll-production heroku maintenance:off --app scihist-digicoll-production
Scale down workers to zero
Possibly disable our hirefire autoscaling to keep it from starting workers back up again
At the appointed time, run the script.
Restore hirefire state