convex-migrations

Zero-downtime Convex schema migrations are automated using a reusable migration runner that applies batched updates to tables, fields, and indices. It tracks progress and handles failures to ensure data consistency without service interruptions.

Renaming fields or changing field types in Convex is handled safely through incremental, code-assisted migration patterns. The process includes backfilling data and tracking batch progress to prevent data loss during the schema evolution.

Batched backfill progress tracking relies on a migration runner pattern utilizing start, update, complete, and fail mutations. This ensures migrations are idempotent and recoverable by monitoring each batch until the schema update finishes successfully.

Yes, failed Convex schema migrations are recoverable. The migration runner uses start, update, complete, and fail mutations to handle failures gracefully, ensuring the migration can resume or complete reliably without losing prior batch progress.

A migration runner is needed when evolving Convex schemas involves complex changes like adding indices, removing fields, or large data backfills. It provides controlled, batched execution to mitigate risks of downtime and inconsistent data.