Sorry for the troubles, and thanks for sharing here. I have forwarded your comment to engineering teams to investigate if related to the migration.
To avoid blockers, I'd also suggest opening a support ticket [0] to let support and billing teams know and escalate. Include this HN comment and my name if you like.
[0] https://support.gitlab.com/hc/en-us/requests/new?ticket_form...
I have talked with the fulfillment team at GitLab [0], and there was a problem identified with workers not syncing the purchased CI minutes after the DB migration. Corrections have been made, and potentially you should see the synced CI minutes soon. Support teams have access and can check with you, I do not have access.
Setting up your own runners is a sensible approach, and allows you to scale them for your own needs too. Maybe this workshop about pipeline efficiency can help with more ideas and insights [1].
[0] https://about.gitlab.com/handbook/engineering/development/fu...
[1] https://about.gitlab.com/handbook/marketing/community-relati...