it seems AWS autoscaling groups has an “instance refresh” functionality, which looks like it would fit the bill

Does that restart the machine or replace it?

it creates new instances

tried it in our dev env that is also production topology, and it worked well

Glad it worked for you. An important thing to remember is that i3.large / i3.xlarge instances support local valcache on the NVMe SSDs. If you replace instances with new ones, you lose a 500 / 1000 GB SSD backed cache of your database (per qg instance) and the new instances will need to rebuild that cache, one miss at a time. Maybe it's not a problem for your situation, but if you find that you're troubleshooting operational issues (degraded query performance, for example), refreshing (aka replacing) the instances shouldn't be the first thing you try. All of the above assumes your QG instances are either i3.large or i3.xlarge.

good to know… not a problem here as this is a temporary measure for something that happens rarely

Do you mean a transaction?

You may be looking for https://docs.datomic.com/cloud/transactions/transaction-processing.html#cancel

yes, a transaction

thank you!