cloudflare-d1

name	cloudflare-d1
description	Build with D1 serverless SQLite database on Cloudflare's edge. Use when: creating databases, writing SQL migrations, querying D1 from Workers, handling relational data, or troubleshooting D1_ERROR, statement too long, migration failures, or query performance issues. Prevents 14 documented errors.
user-invocable	true

Cloudflare D1 Database

Status: Production Ready ✅ Last Updated: 2026-01-20 Dependencies: cloudflare-worker-base (for Worker setup) Latest Versions: wrangler@4.59.2, @cloudflare/workers-types@4.20260109.0

Recent Updates (2025):

• Nov 2025: Jurisdiction support (data localization compliance), remote bindings GA (wrangler@4.37.0+), automatic resource provisioning
• Sept 2025: Automatic read-only query retries (up to 2 attempts), remote bindings public beta
• July 2025: Storage limits increased (250GB → 1TB), alpha backup access removed, REST API 50-500ms faster
• May 2025: HTTP API permissions security fix (D1:Edit required for writes)
• April 2025: Read replication public beta (read-only replicas across regions)
• Feb 2025: PRAGMA optimize support, read-only access permission bug fix
• Jan 2025: Free tier limits enforcement (Feb 10 start), Worker API 40-60% faster queries

---

Quick Start (5 Minutes)

1. Create D1 Database

# Create a new D1 database
npx wrangler d1 create my-database

# Output includes database_id - save this!
# ✅ Successfully created DB 'my-database'
#
# [[d1_databases]]
# binding = "DB"
# database_name = "my-database"
# database_id = "<UUID>"

2. Configure Bindings

Add to your wrangler.jsonc:

{
  "name": "my-worker",
  "main": "src/index.ts",
  "compatibility_date": "2025-10-11",
  "d1_databases": [
    {
      "binding": "DB",                    // Available as env.DB in your Worker
      "database_name": "my-database",      // Name from wrangler d1 create
      "database_id": "<UUID>",             // ID from wrangler d1 create
      "preview_database_id": "local-db"    // For local development
    }
  ]
}

CRITICAL:

• binding is how you access the database in code (env.DB)
• database_id is the production database UUID
• preview_database_id is for local dev (can be any string)
• Never commit real database_id values to public repos - use environment variables or secrets

3. Create Your First Migration

# Create migration file
npx wrangler d1 migrations create my-database create_users_table

# This creates: migrations/0001_create_users_table.sql

Edit the migration file:

-- migrations/0001_create_users_table.sql
DROP TABLE IF EXISTS users;
CREATE TABLE IF NOT EXISTS users (
  user_id INTEGER PRIMARY KEY AUTOINCREMENT,
  email TEXT NOT NULL UNIQUE,
  username TEXT NOT NULL,
  created_at INTEGER NOT NULL,
  updated_at INTEGER
);

-- Create index for common queries
CREATE INDEX IF NOT EXISTS idx_users_email ON users(email);

-- Optimize database
PRAGMA optimize;

4. Apply Migration

# Apply locally first (for testing)
npx wrangler d1 migrations apply my-database --local

# Apply to production when ready
npx wrangler d1 migrations apply my-database --remote

5. Query from Your Worker

// src/index.ts
import { Hono } from 'hono';

type Bindings = {
  DB: D1Database;
};

const app = new Hono<{ Bindings: Bindings }>();

app.get('/api/users/:email', async (c) => {
  const email = c.req.param('email');

  try {
    // ALWAYS use prepared statements with bind()
    const result = await c.env.DB.prepare(
      'SELECT * FROM users WHERE email = ?'
    )
    .bind(email)
    .first();

    if (!result) {
      return c.json({ error: 'User not found' }, 404);
    }

    return c.json(result);
  } catch (error: any) {
    console.error('D1 Error:', error.message);
    return c.json({ error: 'Database error' }, 500);
  }
});

export default app;

---

D1 Migrations System

Migration Workflow

# 1. Create migration
npx wrangler d1 migrations create <DATABASE_NAME> <MIGRATION_NAME>

# 2. List unapplied migrations
npx wrangler d1 migrations list <DATABASE_NAME> --local
npx wrangler d1 migrations list <DATABASE_NAME> --remote

# 3. Apply migrations
npx wrangler d1 migrations apply <DATABASE_NAME> --local   # Test locally
npx wrangler d1 migrations apply <DATABASE_NAME> --remote  # Deploy to production

Migration File Naming

Migrations are automatically versioned:

migrations/
├── 0000_initial_schema.sql
├── 0001_add_users_table.sql
├── 0002_add_posts_table.sql
└── 0003_add_indexes.sql

Rules:

• Files are executed in sequential order
• Each migration runs once (tracked in d1_migrations table)
• Failed migrations roll back (transactional)
• Can't modify or delete applied migrations

Custom Migration Configuration

{
  "d1_databases": [
    {
      "binding": "DB",
      "database_name": "my-database",
      "database_id": "<UUID>",
      "migrations_dir": "db/migrations",        // Custom directory (default: migrations/)
      "migrations_table": "schema_migrations"   // Custom tracking table (default: d1_migrations)
    }
  ]
}

Migration Best Practices

✅ Always Do:

-- Use IF NOT EXISTS to make migrations idempotent
CREATE TABLE IF NOT EXISTS users (...);
CREATE INDEX IF NOT EXISTS idx_users_email ON users(email);

-- Run PRAGMA optimize after schema changes
PRAGMA optimize;

-- Use UPPERCASE BEGIN/END in triggers (lowercase fails remotely)
CREATE TRIGGER update_timestamp
AFTER UPDATE ON users
FOR EACH ROW
BEGIN
  UPDATE users SET updated_at = unixepoch() WHERE user_id = NEW.user_id;
END;

-- Use transactions for data migrations
BEGIN TRANSACTION;
UPDATE users SET updated_at = unixepoch() WHERE updated_at IS NULL;
COMMIT;

❌ Never Do:

-- DON'T include BEGIN TRANSACTION at start of migration file (D1 handles this)
BEGIN TRANSACTION;  -- ❌ Remove this

-- DON'T use lowercase begin/end in triggers (works locally, FAILS remotely)
CREATE TRIGGER my_trigger
AFTER INSERT ON table
begin  -- ❌ Use BEGIN (uppercase)
  UPDATE ...;
end;   -- ❌ Use END (uppercase)

-- DON'T use MySQL/PostgreSQL syntax
ALTER TABLE users MODIFY COLUMN email VARCHAR(255);  -- ❌ Not SQLite

-- DON'T create tables without IF NOT EXISTS
CREATE TABLE users (...);  -- ❌ Fails if table exists

Handling Foreign Keys in Migrations

-- Temporarily disable foreign key checks during schema changes
PRAGMA defer_foreign_keys = true;

-- Make schema changes that would violate foreign keys
ALTER TABLE posts DROP COLUMN author_id;
ALTER TABLE posts ADD COLUMN user_id INTEGER REFERENCES users(user_id);

-- Foreign keys re-enabled automatically at end of migration

---

D1 Workers API

Type Definitions:

interface Env { DB: D1Database; }
type Bindings = { DB: D1Database; };
const app = new Hono<{ Bindings: Bindings }>();

prepare() - PRIMARY METHOD (always use for user input):

const user = await env.DB.prepare('SELECT * FROM users WHERE email = ?')
  .bind(email).first();

Why: Prevents SQL injection, reusable, better performance, type-safe

Query Result Methods:

• .all() → { results, meta } - Get all rows
• .first() → row object or null - Get first row
• .first('column') → value - Get single column value (e.g., COUNT)
• .run() → { success, meta } - Execute INSERT/UPDATE/DELETE (no results)

batch() - CRITICAL FOR PERFORMANCE:

const results = await env.DB.batch([
  env.DB.prepare('SELECT * FROM users WHERE user_id = ?').bind(1),
  env.DB.prepare('SELECT * FROM posts WHERE user_id = ?').bind(1)
]);

• Executes sequentially, single network round trip
• If one fails, remaining statements don't execute
• Use for: bulk inserts, fetching related data

exec() - AVOID IN PRODUCTION:

await env.DB.exec('SELECT * FROM users;'); // Only for migrations/maintenance

• ❌ Never use with user input (SQL injection risk)
• ✅ Only use for: migration files, one-off tasks

---

Query Patterns

Basic CRUD Operations

// CREATE
const { meta } = await env.DB.prepare(
  'INSERT INTO users (email, username, created_at) VALUES (?, ?, ?)'
).bind(email, username, Date.now()).run();
const newUserId = meta.last_row_id;

// READ (single)
const user = await env.DB.prepare('SELECT * FROM users WHERE user_id = ?')
  .bind(userId).first();

// READ (multiple)
const { results } = await env.DB.prepare('SELECT * FROM users LIMIT ?')
  .bind(10).all();

// UPDATE
const { meta } = await env.DB.prepare('UPDATE users SET username = ? WHERE user_id = ?')
  .bind(newUsername, userId).run();
const rowsAffected = meta.rows_written;

// DELETE
await env.DB.prepare('DELETE FROM users WHERE user_id = ?').bind(userId).run();

// COUNT
const count = await env.DB.prepare('SELECT COUNT(*) as total FROM users').first('total');

// EXISTS check
const exists = await env.DB.prepare('SELECT 1 FROM users WHERE email = ? LIMIT 1')
  .bind(email).first();

Pagination Pattern

const page = parseInt(c.req.query('page') || '1');
const limit = 20;
const offset = (page - 1) * limit;

const [countResult, usersResult] = await c.env.DB.batch([
  c.env.DB.prepare('SELECT COUNT(*) as total FROM users'),
  c.env.DB.prepare('SELECT * FROM users ORDER BY created_at DESC LIMIT ? OFFSET ?')
    .bind(limit, offset)
]);

return c.json({
  users: usersResult.results,
  pagination: { page, limit, total: countResult.results[0].total }
});

Batch Pattern (Pseudo-Transactions)

// D1 doesn't support multi-statement transactions, but batch() provides sequential execution
await env.DB.batch([
  env.DB.prepare('UPDATE users SET credits = credits - ? WHERE user_id = ?').bind(amount, fromUserId),
  env.DB.prepare('UPDATE users SET credits = credits + ? WHERE user_id = ?').bind(amount, toUserId),
  env.DB.prepare('INSERT INTO transactions (from_user, to_user, amount) VALUES (?, ?, ?)').bind(fromUserId, toUserId, amount)
]);
// If any statement fails, batch stops (transaction-like behavior)

---

Error Handling

Common Error Types:

• D1_ERROR - General D1 error (often transient)
• D1_EXEC_ERROR - SQL syntax error or limitations
• D1_TYPE_ERROR - Type mismatch (undefined instead of null)
• D1_COLUMN_NOTFOUND - Column doesn't exist

Common Errors and Fixes:

Error	Cause	Solution
Statement too long	Large INSERT with 1000+ rows	Break into batches of 100-250 using batch()
Network connection lost	Transient failure or large import	Implement retry logic (see below) or break into smaller chunks
Too many requests queued	Individual queries in loop	Use batch() instead of loop
D1_TYPE_ERROR	Using undefined in bind	Use null for optional values: .bind(email, bio || null)
Transaction conflicts	BEGIN TRANSACTION in migration	Remove BEGIN/COMMIT (D1 handles automatically)
Foreign key violations	Schema changes break constraints	Use PRAGMA defer_foreign_keys = true
D1_EXEC_ERROR: incomplete input	Multi-line SQL in D1Database.exec()	Use prepared statements or external .sql files (Issue #9133)

Transient Errors Are Expected Behavior

CRITICAL: D1 queries fail transiently with errors like "Network connection lost", "storage operation exceeded timeout", or "isolate exceeded its memory limit". Cloudflare documentation states "a handful of errors every several hours is not unexpected" and recommends implementing retry logic. (D1 FAQ)

Common Transient Errors:

• D1_ERROR: Network connection lost
• D1 DB storage operation exceeded timeout which caused object to be reset
• Internal error while starting up D1 DB storage caused object to be reset
• D1 DB's isolate exceeded its memory limit and was reset

Retry Pattern (Recommended):

async function queryWithRetry<T>(
  fn: () => Promise<T>,
  maxRetries = 3,
  baseDelay = 100
): Promise<T> {
  for (let i = 0; i < maxRetries; i++) {
    try {
      return await fn();
    } catch (error: any) {
      const isTransient = error.message?.includes('Network connection lost') ||
                         error.message?.includes('exceeded timeout') ||
                         error.message?.includes('exceeded its memory limit');

      if (!isTransient || i === maxRetries - 1) throw error;

      // Exponential backoff
      await new Promise(r => setTimeout(r, baseDelay * Math.pow(2, i)));
    }
  }
  throw new Error('Max retries exceeded');
}

// Usage
const user = await queryWithRetry(() =>
  env.DB.prepare('SELECT * FROM users WHERE email = ?').bind(email).first()
);

Automatic Retries (Sept 2025): D1 automatically retries read-only queries (SELECT, EXPLAIN, WITH) up to 2 times on retryable errors. Check meta.total_attempts in response for retry count. Write queries should still implement custom retry logic.

---

Performance Optimization

Index Best Practices:

• ✅ Index columns in WHERE clauses: CREATE INDEX idx_users_email ON users(email)
• ✅ Index foreign keys: CREATE INDEX idx_posts_user_id ON posts(user_id)
• ✅ Index columns for sorting: CREATE INDEX idx_posts_created_at ON posts(created_at DESC)
• ✅ Multi-column indexes: CREATE INDEX idx_posts_user_published ON posts(user_id, published)
• ✅ Partial indexes: CREATE INDEX idx_users_active ON users(email) WHERE deleted = 0
• ✅ Test with: EXPLAIN QUERY PLAN SELECT ...

PRAGMA optimize (Feb 2025):

CREATE INDEX idx_users_email ON users(email);
PRAGMA optimize;  -- Run after schema changes

Query Optimization:

• ✅ Use specific columns (not SELECT *)
• ✅ Always include LIMIT on large result sets
• ✅ Use indexes for WHERE conditions
• ❌ Avoid functions in WHERE (can't use indexes): WHERE LOWER(email) → store lowercase instead

---

Local Development

Local vs Remote (Nov 2025 - Remote Bindings GA):

# Local database (automatic creation)
npx wrangler d1 migrations apply my-database --local
npx wrangler d1 execute my-database --local --command "SELECT * FROM users"

# Remote database
npx wrangler d1 execute my-database --remote --command "SELECT * FROM users"

# Remote bindings (wrangler@4.37.0+) - connect local Worker to deployed D1
# Add to wrangler.jsonc: { "binding": "DB", "remote": true }

Remote Bindings Connection Timeout

Known Issue: When using remote D1 bindings ({ "remote": true }), the connection times out after exactly 1 hour of inactivity. (GitHub Issue #10801)

Error: D1_ERROR: Failed to parse body as JSON, got: error code: 1031

Workaround:

// Keep connection alive with periodic query (optional)
setInterval(async () => {
  try {
    await env.DB.prepare('SELECT 1').first();
  } catch (e) {
    console.log('Connection keepalive failed:', e);
  }
}, 30 * 60 * 1000); // Every 30 minutes

Or simply restart your dev server if queries fail after 1 hour of inactivity.

Multi-Worker Development (Service Bindings)

When running multiple Workers with service bindings in a single wrangler dev process, the auxiliary worker cannot access its D1 binding because both workers share the same persistence path. (GitHub Issue #11121)

Solution: Use --persist-to flag to point all workers to the same persistence store:

# Apply worker2 migrations to worker1's persistence path
cd worker2
npx wrangler d1 migrations apply DB --local --persist-to=../worker1/.wrangler/state

# Now both workers can access D1
cd ../worker1
npx wrangler dev  # Both workers share the same D1 data

Local Database Location: .wrangler/state/v3/d1/miniflare-D1DatabaseObject/<database_id>.sqlite

Seed Local Database:

npx wrangler d1 execute my-database --local --file=seed.sql

---

Scaling & Limitations

10 GB Database Size Limit - Sharding Pattern

D1 has a hard 10 GB per database limit, but Cloudflare supports up to 50,000 databases per Worker. Use sharding to scale beyond 10 GB. (DEV.to Article)

Hash-based sharding example (10 databases = 100 GB capacity):

// Hash user ID to shard number
function getShardId(userId: string): number {
  const hash = Array.from(userId).reduce((acc, char) =>
    ((acc << 5) - acc) + char.charCodeAt(0), 0
  );
  return Math.abs(hash) % 10; // 10 shards
}

// wrangler.jsonc - Define 10 database shards
{
  "d1_databases": [
    { "binding": "DB_SHARD_0", "database_id": "..." },
    { "binding": "DB_SHARD_1", "database_id": "..." },
    { "binding": "DB_SHARD_2", "database_id": "..." },
    // ... up to DB_SHARD_9
  ]
}

// Get correct shard for user
function getUserDb(env: Env, userId: string): D1Database {
  const shardId = getShardId(userId);
  return env[`DB_SHARD_${shardId}`];
}

// Query user's data from correct shard
const db = getUserDb(env, userId);
const user = await db.prepare('SELECT * FROM users WHERE user_id = ?')
  .bind(userId).first();

Alternative: Tenant-based sharding (one database per customer/tenant)

2 MB Row Size Limit - Hybrid D1 + R2 Pattern

D1 has a 2 MB row size limit. For large content (HTML, JSON, images), use R2 for storage and D1 for metadata. (DEV.to Article)

Error: database row size exceeded maximum allowed size

Solution - Hybrid storage pattern:

// 1. Store large content in R2
const contentKey = `pages/${crypto.randomUUID()}.html`;
await env.R2_BUCKET.put(contentKey, largeHtmlContent);

// 2. Store metadata in D1
await env.DB.prepare(`
  INSERT INTO pages (url, r2_key, size, created_at)
  VALUES (?, ?, ?, ?)
`).bind(url, contentKey, largeHtmlContent.length, Date.now()).run();

// 3. Retrieve content
const page = await env.DB.prepare('SELECT * FROM pages WHERE url = ?')
  .bind(url).first();

if (page) {
  const content = await env.R2_BUCKET.get(page.r2_key);
  const html = await content.text();
}

Database Portability - PostgreSQL Migration Considerations

If you plan to migrate from D1 (SQLite) to Hyperdrive (PostgreSQL) later, use consistent lowercase naming. PostgreSQL is case-sensitive for table and column names, while SQLite is not. (Mats' Blog)

-- Use lowercase for portability
CREATE TABLE users (user_id INTEGER, email TEXT);
CREATE INDEX idx_users_email ON users(email);

-- NOT: CREATE TABLE Users (UserId INTEGER, Email TEXT);

FTS5 Full-Text Search

Case Sensitivity: Always use lowercase "fts5" when creating virtual tables. Uppercase may cause "not authorized" errors. (Cloudflare Community)

-- Correct
CREATE VIRTUAL TABLE search_index USING fts5(
  title,
  content,
  tokenize = 'porter unicode61'
);

-- Query the index
SELECT * FROM search_index WHERE search_index MATCH 'query terms';

Export Limitation: Databases with FTS5 virtual tables cannot be exported using wrangler d1 export. Drop virtual tables before export, then recreate after import. (GitHub Issue #9519)

Large Import/Export Operations

Network Timeout on Large Imports: Files with 5000+ INSERT statements may fail with "Network connection lost" error. (GitHub Issue #11958)

Solutions:

1. Break large files into smaller chunks (<5000 statements per file)
2. Use batch() API from Worker instead of wrangler CLI
3. Import to local first, then use Time Travel to restore to remote
4. Reduce individual statement size (100-250 rows per INSERT)

Windows-Specific Issue: On Windows 11, large SQL files exported from D1 may fail to re-import with "HashIndex detected hash table inconsistency". (GitHub Issue #11708)

Workaround: Delete .wrangler directory before executing:

rm -rf .wrangler
npx wrangler d1 execute db-name --file=database.sql

---

Best Practices Summary

✅ Always Do:

1. Use prepared statements with .bind() for user input
2. Use .batch() for multiple queries (reduces latency)
3. Create indexes on frequently queried columns
4. Run PRAGMA optimize after schema changes
5. Use IF NOT EXISTS in migrations for idempotency
6. Test migrations locally before applying to production
7. Handle errors gracefully with try/catch
8. Use null instead of undefined for optional values
9. Validate input before binding to queries
10. Check meta.rows_written after UPDATE/DELETE

❌ Never Do:

1. Never use .exec() with user input (SQL injection risk)
2. Never hardcode database_id in public repos
3. Never use undefined in bind parameters (causes D1_TYPE_ERROR)
4. Never fire individual queries in loops (use batch instead)
5. Never forget LIMIT on potentially large result sets
6. Never use SELECT * in production (specify columns)
7. Never include BEGIN TRANSACTION in migration files
8. Never modify applied migrations (create new ones)
9. Never skip error handling on database operations
10. Never assume queries succeed (always check results)

---

Known Issues Prevented

This skill prevents 14 documented D1 errors:

Issue #	Error/Issue	Description	How to Avoid	Source
#1	Statement too long	Large INSERT statements exceed D1 limits	Break into batches of 100-250 rows using batch()	Existing
#2	Transaction conflicts	BEGIN TRANSACTION in migration files	Remove BEGIN/COMMIT (D1 handles automatically)	Existing
#3	Foreign key violations	Schema changes break foreign key constraints	Use PRAGMA defer_foreign_keys = true in migrations	Existing
#4	Rate limiting / queue overload	Too many individual queries	Use batch() instead of loops	Existing
#5	Memory limit exceeded	Query loads too much data into memory	Add LIMIT, paginate results, shard queries	Existing
#6	Type mismatch errors	Using undefined instead of null in bind()	Always use null for optional values	Existing
#7	Lowercase BEGIN in triggers	Triggers with lowercase begin/end fail remotely	Use uppercase BEGIN/END keywords (Issue #10998)	TIER 1
#8	Remote bindings timeout	Connection times out after 1 hour of inactivity	Restart dev server or implement keepalive pattern (Issue #10801)	TIER 1
#9	Service bindings D1 access	Auxiliary worker can't access D1 in multi-worker dev	Use --persist-to flag to share persistence path (Issue #11121)	TIER 1
#10	Transient network errors	Random "Network connection lost" failures	Implement exponential backoff retry logic (D1 FAQ)	TIER 1
#11	FTS5 breaks export	Databases with FTS5 virtual tables can't export	Drop virtual tables before export, recreate after import (Issue #9519)	TIER 1
#12	Multi-line SQL in exec()	D1Database.exec() fails on multi-line SQL	Use prepared statements or external .sql files (Issue #9133)	TIER 1
#13	10 GB database limit	Single database limited to 10 GB	Implement sharding across multiple databases (Community)	TIER 2
#14	2 MB row size limit	Rows exceeding 2 MB fail	Use hybrid D1 + R2 storage pattern (Community)	TIER 2

---

Wrangler Commands Reference

# Database management
wrangler d1 create <DATABASE_NAME>
wrangler d1 list
wrangler d1 delete <DATABASE_NAME>
wrangler d1 info <DATABASE_NAME>

# Migrations
wrangler d1 migrations create <DATABASE_NAME> <MIGRATION_NAME>
wrangler d1 migrations list <DATABASE_NAME> --local|--remote
wrangler d1 migrations apply <DATABASE_NAME> --local|--remote

# Execute queries
wrangler d1 execute <DATABASE_NAME> --local|--remote --command "SELECT * FROM users"
wrangler d1 execute <DATABASE_NAME> --local|--remote --file=./query.sql

# Time Travel (view historical data)
wrangler d1 time-travel info <DATABASE_NAME> --timestamp "2025-10-20"
wrangler d1 time-travel restore <DATABASE_NAME> --timestamp "2025-10-20"

---

Official Documentation

• D1 Overview: https://developers.cloudflare.com/d1/
• Get Started: https://developers.cloudflare.com/d1/get-started/
• Migrations: https://developers.cloudflare.com/d1/reference/migrations/
• Workers API: https://developers.cloudflare.com/d1/worker-api/
• Best Practices: https://developers.cloudflare.com/d1/best-practices/
• Wrangler Commands: https://developers.cloudflare.com/workers/wrangler/commands/#d1

---

Ready to build with D1! 🚀

---

Last verified: 2026-01-20 | Skill version: 3.0.0 | Changes: Added 8 new known issues from community research (TIER 1-2 findings): trigger case sensitivity, remote binding timeouts, multi-worker dev patterns, transient error handling, FTS5 limitations, sharding patterns, hybrid D1+R2 storage, and database portability considerations.