Convex Migrations
Evolve your Convex database schema safely with patterns for adding fields, backfilling data, removing deprecated fields, and maintaining zero-downtime deployments.
Documentation Sources
Before implementing, do not assume; fetch the latest documentation:
- Primary: https://docs.convex.dev/database/schemas
- Schema Overview: https://docs.convex.dev/database
- Migration Patterns: https://stack.convex.dev/migrate-data-postgres-to-convex
- For broader context: https://docs.convex.dev/llms.txt
Instructions
Migration Philosophy
Convex handles schema evolution differently than traditional databases:
- No explicit migration files or commands
- Schema changes deploy instantly with
npx convex dev - Existing data is not automatically transformed
- Use optional fields and backfill mutations for safe migrations
Adding New Fields
Start with optional fields, then backfill:
// Step 1: Add optional field to schema
// convex/schema.ts
import { defineSchema, defineTable } from "convex/server";
import { v } from "convex/values";
export default defineSchema({
users: defineTable({
name: v.string(),
email: v.string(),
// New field - start as optional
avatarUrl: v.optional(v.string()),
}),
});
// Step 2: Update code to handle both cases
// convex/users.ts
import { query } from "./_generated/server";
import { v } from "convex/values";
export const getUser = query({
args: { userId: v.id("users") },
returns: v.union(
v.object({
_id: v.id("users"),
name: v.string(),
email: v.string(),
avatarUrl: v.union(v.string(), v.null()),
}),
v.null()
),
handler: async (ctx, args) => {
const user = await ctx.db.get(args.userId);
if (!user) return null;
return {
_id: user._id,
name: user.name,
email: user.email,
// Handle missing field gracefully
avatarUrl: user.avatarUrl ?? null,
};
},
});
// Step 3: Backfill existing documents
// convex/migrations.ts
import { internalMutation } from "./_generated/server";
import { internal } from "./_generated/api";
import { v } from "convex/values";
const BATCH_SIZE = 100;
export const backfillAvatarUrl = internalMutation({
args: {
cursor: v.optional(v.string()),
},
returns: v.object({
processed: v.number(),
hasMore: v.boolean(),
}),
handler: async (ctx, args) => {
const result = await ctx.db
.query("users")
.paginate({ numItems: BATCH_SIZE, cursor: args.cursor ?? null });
let processed = 0;
for (const user of result.page) {
// Only update if field is missing
if (user.avatarUrl === undefined) {
await ctx.db.patch(user._id, {
avatarUrl: generateDefaultAvatar(user.name),
});
processed++;
}
}
// Schedule next batch if needed
if (!result.isDone) {
await ctx.scheduler.runAfter(0, internal.migrations.backfillAvatarUrl, {
cursor: result.continueCursor,
});
}
return {
processed,
hasMore: !result.isDone,
};
},
});
function generateDefaultAvatar(name: string): string {
return `https://api.dicebear.com/7.x/initials/svg?seed=${encodeURIComponent(name)}`;
}
// Step 4: After backfill completes, make field required
// convex/schema.ts
export default defineSchema({
users: defineTable({
name: v.string(),
email: v.string(),
avatarUrl: v.string(), // Now required
}),
});
Removing Fields
Remove field usage before removing from schema:
// Step 1: Stop using the field in queries and mutations
// Mark as deprecated in code comments
// Step 2: Remove field from schema (make optional first if needed)
// convex/schema.ts
export default defineSchema({
posts: defineTable({
title: v.string(),
content: v.string(),
authorId: v.id("users"),
// legacyField: v.optional(v.string()), // Remove this line
}),
});
// Step 3: Optionally clean up existing data
// convex/migrations.ts
export const removeDeprecatedField = internalMutation({
args: {
cursor: v.optional(v.string()),
},
returns: v.null(),
handler: async (ctx, args) => {
const result = await ctx.db
.query("posts")
.paginate({ numItems: 100, cursor: args.cursor ?? null });
for (const post of result.page) {
// Use replace to remove the field entirely
const { legacyField, ...rest } = post as typeof post & { legacyField?: string };
if (legacyField !== undefined) {
await ctx.db.replace(post._id, rest);
}
}
if (!result.isDone) {
await ctx.scheduler.runAfter(0, internal.migrations.removeDeprecatedField, {
cursor: result.continueCursor,
});
}
return null;
},
});
Renaming Fields
Renaming requires copying data to new field, then removing old:
// Step 1: Add new field as optional
// convex/schema.ts
export default defineSchema({
users: defineTable({
userName: v.string(), // Old field
displayName: v.optional(v.string()), // New field
}),
});
// Step 2: Update code to read from new field with fallback
export const getUser = query({
args: { userId: v.id("users") },
returns: v.object({
_id: v.id("users"),
displayName: v.string(),
}),
handler: async (ctx, args) => {
const user = await ctx.db.get(args.userId);
if (!user) throw new Error("User not found");
return {
_id: user._id,
// Read new field, fall back to old
displayName: user.displayName ?? user.userName,
};
},
});
// Step 3: Backfill to copy data
export const backfillDisplayName = internalMutation({
args: { cursor: v.optional(v.string()) },
returns: v.null(),
handler: async (ctx, args) => {
const result = await ctx.db
.query("users")
.paginate({ numItems: 100, cursor: args.cursor ?? null });
for (const user of result.page) {
if (user.displayName === undefined) {
await ctx.db.patch(user._id, {
displayName: user.userName,
});
}
}
if (!result.isDone) {
await ctx.scheduler.runAfter(0, internal.migrations.backfillDisplayName, {
cursor: result.continueCursor,
});
}
return null;
},
});
// Step 4: After backfill, update schema to make new field required
// and remove old field
export default defineSchema({
users: defineTable({
// userName removed
displayName: v.string(),
}),
});
Adding Indexes
Add indexes before using them in queries:
// Step 1: Add index to schema
// convex/schema.ts
export default defineSchema({
posts: defineTable({
title: v.string(),
authorId: v.id("users"),
publishedAt: v.optional(v.number()),
status: v.string(),
})
.index("by_author", ["authorId"])
// New index
.index("by_status_and_published", ["status", "publishedAt"]),
});
// Step 2: Deploy schema change
// Run: npx convex dev
// Step 3: Now use the index in queries
export const getPublishedPosts = query({
args: {},
returns: v.array(v.object({
_id: v.id("posts"),
title: v.string(),
publishedAt: v.number(),
})),
handler: async (ctx) => {
const posts = await ctx.db
.query("posts")
.withIndex("by_status_and_published", (q) =>
q.eq("status", "published")
)
.order("desc")
.take(10);
return posts
.filter((p) => p.publishedAt !== undefined)
.map((p) => ({
_id: p._id,
title: p.title,
publishedAt: p.publishedAt!,
}));
},
});
Changing Field Types
Type changes require careful migration:
// Example: Change from string to number for a "priority" field
// Step 1: Add new field with new type
// convex/schema.ts
export default defineSchema({
tasks: defineTable({
title: v.string(),
priority: v.string(), // Old: "low", "medium", "high"
priorityLevel: v.optional(v.number()), // Ne