Snapshots and migrations
A microfeed snapshot packages the D1 schema and durable data together with every object in the production R2 bucket. The archive also records checksums and the ordered migration history needed to validate a restore.
Create a backup
Section titled “Create a backup”yarn manage snapshot create \ --instance <source-name> \ --output <new-backup-file>.tar.gzExpected result: a new archive at the requested path. Existing output files are not overwritten. Store the archive like private site data; it can contain unpublished content and media.
Test a production snapshot locally
Section titled “Test a production snapshot locally”yarn manage snapshot pull \ --instance <source-name> \ --local-instance <new-local-name>This downloads a temporary snapshot, restores it into a new local instance, and removes the temporary archive. It does not change production.
Restore remotely
Section titled “Restore remotely”Remote restore replaces data and therefore has stricter requirements. The
target must be a newly initialized, unchanged deployment. First run the exact
restore with --dry-run, inspect the source and target, then use
--confirm <target-instance-name> only when every identifier matches.
See the canonical snapshot reference for eligibility checks, resumable maintenance state, and all options.
Migrate an older Pages installation
Section titled “Migrate an older Pages installation”Older microfeed deployments hosted with Cloudflare Pages use:
yarn manage migrate-pagesThis command will create a new Workers project to connect to the existing d1 database and r2 bucket. So the old Pages project and the new Workers project both connect to the same d1 database and r2 bucket. You can decide when to switch the custom domain name from the old Pages project to the new Workers project, and vice versa.
See the canonical migrate-pages reference for all options.

