Using Litestream as a Go Library

Litestream can be embedded directly into your Go application as a library, giving you programmatic control over database replication without running a separate process. This approach is useful when you need tighter integration with your application lifecycle or want to customize replication behavior.

⚠️

The Litestream library API is not considered stable and may change between versions. The CLI interface is recommended for production use cases where API stability is important.

When to use the library

Embed replication directly in your Go application without managing a separate Litestream process.
Implement custom restore-on-startup patterns for ephemeral or containerized deployments.
Integrate replication lifecycle with your application’s graceful shutdown handling.
Manage multiple databases from a single application with unified compaction.
Avoid when you need a stable, well-documented interface—the CLI is more mature.

Prerequisites

Go 1.24 or later.
A SQLite driver. The examples use modernc.org/sqlite (pure Go, no CGO required).
Familiarity with Litestream concepts from the Getting Started guide.

Installation

Add Litestream to your Go module:

go get github.com/benbjohnson/litestream

Import the packages you need:

import (
    "github.com/benbjohnson/litestream"
    "github.com/benbjohnson/litestream/file"  // For file backend
    "github.com/benbjohnson/litestream/s3"    // For S3 backend
)

Important constraints

When using Litestream as a library, be aware of these critical requirements:

Required SQLite driver: You must use modernc.org/sqlite. Litestream uses this driver internally, and mixing drivers causes lock conflicts on POSIX systems (Linux, macOS, BSD) due to per-process SQLite locks.
Lifecycle management: You cannot call litestream.DB.Close() or Replica.Stop(true) while your application still has open database connections. Either close all your app’s database connections first, or only close Litestream when your process is shutting down.
PRAGMA configuration: Use DSN parameters (e.g., ?_pragma=busy_timeout(5000)) instead of PRAGMA statements via ExecContext. An sql.DB is a connection pool, and ExecContext only applies the PRAGMA to one random connection from the pool.

Core concepts

The library uses three main types:

DB: Wraps a SQLite database path and manages its replica.
Replica: Handles replication to a specific backend (S3, file, etc.).
Store: Manages one or more databases and runs background compaction.

The typical lifecycle is:

Create a DB wrapper for your SQLite database path.
Create a replica client for your storage backend.
Attach the replica to the database.
Create a Store with compaction levels.
Open the store (starts replication and compaction monitors).
Open your application’s SQLite connection separately.
Close the store on shutdown.

Basic usage

This example replicates a SQLite database to the local filesystem:

package main

import (
    "context"
    "database/sql"
    "fmt"
    "log"
    "time"

    _ "modernc.org/sqlite"

    "github.com/benbjohnson/litestream"
    "github.com/benbjohnson/litestream/file"
)

func main() {
    ctx := context.Background()
    dbPath := "./myapp.db"
    replicaPath := "./replica"

    // 1. Create the Litestream DB wrapper
    db := litestream.NewDB(dbPath)

    // 2. Create a replica client (file-based)
    client := file.NewReplicaClient(replicaPath)

    // 3. Create a replica and attach it to the database
    replica := litestream.NewReplicaWithClient(db, client)
    db.Replica = replica
    client.Replica = replica // Preserves file permissions

    // 4. Create compaction levels (L0 required, plus at least one more)
    levels := litestream.CompactionLevels{
        {Level: 0},
        {Level: 1, Interval: 10 * time.Second},
    }

    // 5. Create a Store to manage the database and background compaction
    store := litestream.NewStore([]*litestream.DB{db}, levels)

    // 6. Open the store (starts replication and compaction monitors)
    if err := store.Open(ctx); err != nil {
        log.Fatalf("open store: %v", err)
    }
    defer func() {
        if err := store.Close(context.Background()); err != nil {
            log.Printf("close store: %v", err)
        }
    }()

    // 7. Open your app's SQLite connection separately
    // Use DSN parameters for PRAGMAs to ensure they apply to all pooled connections
    dsn := fmt.Sprintf("file:%s?_pragma=busy_timeout(5000)&_pragma=journal_mode(wal)", dbPath)
    sqlDB, err := sql.Open("sqlite", dsn)
    if err != nil {
        log.Fatalf("open sqlite: %v", err)
    }
    defer sqlDB.Close()

    // Use sqlDB for your application queries
    fmt.Println("Replication started. Database ready for use.")
}

S3 backend with restore-on-startup

For production deployments, you typically want to restore from backup if the local database doesn’t exist. This pattern works well with ephemeral containers:

package main

import (
    "context"
    "database/sql"
    "errors"
    "fmt"
    "log"
    "os"
    "time"

    _ "modernc.org/sqlite"

    "github.com/benbjohnson/litestream"
    "github.com/benbjohnson/litestream/s3"
)

const dbPath = "./myapp.db"

func main() {
    ctx := context.Background()

    // Configure S3 client
    client := s3.NewReplicaClient()
    client.Bucket = os.Getenv("LITESTREAM_BUCKET")
    client.Path = os.Getenv("LITESTREAM_PATH")
    client.Region = os.Getenv("AWS_REGION")
    client.AccessKeyID = os.Getenv("AWS_ACCESS_KEY_ID")
    client.SecretAccessKey = os.Getenv("AWS_SECRET_ACCESS_KEY")

    // Restore from S3 if local database doesn't exist
    if err := restoreIfNotExists(ctx, client, dbPath); err != nil {
        log.Fatalf("restore: %v", err)
    }

    // Set up replication (same pattern as basic example)
    db := litestream.NewDB(dbPath)
    replica := litestream.NewReplicaWithClient(db, client)
    db.Replica = replica

    levels := litestream.CompactionLevels{
        {Level: 0},
        {Level: 1, Interval: 10 * time.Second},
    }
    store := litestream.NewStore([]*litestream.DB{db}, levels)

    if err := store.Open(ctx); err != nil {
        log.Fatalf("open store: %v", err)
    }
    defer store.Close(context.Background())

    // Open app database with DSN parameters for PRAGMAs
    dsn := fmt.Sprintf("file:%s?_pragma=busy_timeout(5000)&_pragma=journal_mode(wal)", dbPath)
    sqlDB, _ := sql.Open("sqlite", dsn)
    defer sqlDB.Close()

    log.Println("Database ready with S3 replication")
}

func restoreIfNotExists(ctx context.Context, client *s3.ReplicaClient, dbPath string) error {
    // Check if database already exists
    if _, err := os.Stat(dbPath); err == nil {
        log.Println("Local database found, skipping restore")
        return nil
    } else if !os.IsNotExist(err) {
        return err
    }

    log.Println("Local database not found, attempting restore from S3...")

    // Initialize the client
    if err := client.Init(ctx); err != nil {
        return fmt.Errorf("init s3 client: %w", err)
    }

    // Create a replica (without DB) for restore
    replica := litestream.NewReplicaWithClient(nil, client)

    // Set up restore options
    opt := litestream.NewRestoreOptions()
    opt.OutputPath = dbPath

    // Attempt restore
    if err := replica.Restore(ctx, opt); err != nil {
        // No backup exists - that's OK for a fresh deployment
        if errors.Is(err, litestream.ErrTxNotAvailable) ||
           errors.Is(err, litestream.ErrNoSnapshots) {
            log.Println("No backup found, will create new database")
            return nil
        }
        return err
    }

    log.Println("Database restored from S3")
    return nil
}

Set the required environment variables:

export LITESTREAM_BUCKET="my-backup-bucket"
export LITESTREAM_PATH="databases/myapp"
export AWS_REGION="us-east-1"
export AWS_ACCESS_KEY_ID="AKIAxxxx"
export AWS_SECRET_ACCESS_KEY="xxxx"

Supported backends

The library supports all Litestream replica backends:

Backend	Client Package	URL Scheme
Local filesystem	`github.com/benbjohnson/litestream/file`	`file://`
Amazon S3	`github.com/benbjohnson/litestream/s3`	`s3://`
Google Cloud Storage	`github.com/benbjohnson/litestream/gcs`	`gs://`
Azure Blob Storage	`github.com/benbjohnson/litestream/abs`	`abs://`
Alibaba Cloud OSS	`github.com/benbjohnson/litestream/oss`	`oss://`
SFTP	`github.com/benbjohnson/litestream/sftp`	`sftp://`
NATS JetStream	`github.com/benbjohnson/litestream/nats`	`nats://`
WebDAV	`github.com/benbjohnson/litestream/webdav`	`webdav://`

You can also create a client from a URL:

client, err := litestream.NewReplicaClientFromURL("s3://my-bucket/path")

Configuration options

Compaction levels

Compaction reduces the number of files in your replica by merging smaller files:

levels := litestream.CompactionLevels{
    {Level: 0},                                    // Required: L0 files
    {Level: 1, Interval: 10 * time.Second},        // Compact every 10s
    {Level: 2, Interval: 1 * time.Minute},         // Optional: longer retention
}

Database options

Configure monitoring and checkpointing behavior:

db := litestream.NewDB(dbPath)
db.MonitorInterval = 1 * time.Second      // How often to check for changes
db.CheckpointInterval = 30 * time.Second  // WAL checkpoint frequency
db.MinCheckpointPageN = 1000              // Minimum pages before checkpoint

Replica options

Configure sync behavior:

replica := litestream.NewReplicaWithClient(db, client)
replica.SyncInterval = 1 * time.Second  // How often to sync to replica

Graceful shutdown

Always close the store before your application exits to ensure all pending writes are flushed to the replica:

// Handle shutdown signals
sigCh := make(chan os.Signal, 1)
signal.Notify(sigCh, syscall.SIGINT, syscall.SIGTERM)

<-sigCh
log.Println("Shutting down...")

// Close the store with a timeout context if needed
ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
defer cancel()

if err := store.Close(ctx); err != nil {
    log.Printf("close store: %v", err)
}

Troubleshooting

“database is locked” errors: Ensure you set busy_timeout via DSN parameters (e.g., ?_pragma=busy_timeout(5000)). A value of 5000ms (5 seconds) works well. Using ExecContext with PRAGMA statements only affects one connection in the pool.
WAL mode not enabled: Set journal_mode via DSN parameters (e.g., ?_pragma=journal_mode(wal)). Litestream requires WAL mode for replication.
No replication occurring: Verify the store is opened successfully and that your replica client has valid credentials.
Restore fails with “no snapshots”: This is expected for a fresh deployment with no prior backups. Handle ErrNoSnapshots gracefully.

Using Litestream as a Go Library

When to use the library#

Prerequisites#

Installation#

Important constraints#

Core concepts#

Basic usage#

S3 backend with restore-on-startup#

Supported backends#

Configuration options#

Compaction levels#

Database options#

Replica options#

Graceful shutdown#

Troubleshooting#

See also#