Backup (#71)

backup.go

@@ -0,0 +1,80 @@
+package pogreb
+
+import (
+	"io"
+	"os"
+
+	"github.com/akrylysov/pogreb/fs"
+)
+
+func touchFile(fsys fs.FileSystem, path string) error {
+	f, err := fsys.OpenFile(path, os.O_CREATE|os.O_TRUNC, os.FileMode(0640))
+	if err != nil {
+		return err
+	}
+	return f.Close()
+}
+
+// Backup creates a database backup at the specified path.
+func (db *DB) Backup(path string) error {
+	// Make sure the compaction is not running during backup.
+	db.maintenanceMu.Lock()
+	defer db.maintenanceMu.Unlock()
+
+	if err := db.opts.rootFS.MkdirAll(path, 0755); err != nil {
+		return err
+	}
+
+	db.mu.RLock()
+	var segments []*segment
+	activeSegmentSizes := make(map[uint16]int64)
+	for _, seg := range db.datalog.segmentsBySequenceID() {
+		segments = append(segments, seg)
+		if !seg.meta.Full {
+			// Save the size of the active segments to copy only the data persisted up to the point
+			// of when the backup started.
+			activeSegmentSizes[seg.id] = seg.size
+		}
+	}
+	db.mu.RUnlock()
+
+	srcFS := db.opts.FileSystem
+	dstFS := fs.Sub(db.opts.rootFS, path)
+
+	for _, seg := range segments {
+		name := segmentName(seg.id, seg.sequenceID)
+		mode := os.FileMode(0640)
+		srcFile, err := srcFS.OpenFile(name, os.O_RDONLY, mode)
+		if err != nil {
+			return err
+		}
+
+		dstFile, err := dstFS.OpenFile(name, os.O_CREATE|os.O_RDWR|os.O_TRUNC, mode)
+		if err != nil {
+			return err
+		}
+
+		if srcSize, ok := activeSegmentSizes[seg.id]; ok {
+			if _, err := io.CopyN(dstFile, srcFile, srcSize); err != nil {
+				return err
+			}
+		} else {
+			if _, err := io.Copy(dstFile, srcFile); err != nil {
+				return err
+			}
+		}
+
+		if err := srcFile.Close(); err != nil {
+			return err
+		}
+		if err := dstFile.Close(); err != nil {
+			return err
+		}
+	}
+
+	if err := touchFile(dstFS, lockName); err != nil {
+		return err
+	}
+
+	return nil
+}

backup_test.go

@@ -0,0 +1,63 @@
+package pogreb
+
+import (
+	"testing"
+
+	"github.com/akrylysov/pogreb/internal/assert"
+)
+
+const testDBBackupName = testDBName + ".backup"
+
+func TestBackup(t *testing.T) {
+	opts := &Options{
+		maxSegmentSize:             1024,
+		compactionMinSegmentSize:   520,
+		compactionMinFragmentation: 0.02,
+	}
+
+	run := func(name string, f func(t *testing.T, db *DB)) bool {
+		return t.Run(name, func(t *testing.T) {
+			db, err := createTestDB(opts)
+			assert.Nil(t, err)
+			f(t, db)
+			assert.Nil(t, db.Close())
+			_ = cleanDir(testDBBackupName)
+		})
+	}
+
+	run("empty", func(t *testing.T, db *DB) {
+		assert.Nil(t, db.Backup(testDBBackupName))
+		db2, err := Open(testDBBackupName, opts)
+		assert.Nil(t, err)
+		assert.Nil(t, db2.Close())
+	})
+
+	run("single segment", func(t *testing.T, db *DB) {
+		assert.Nil(t, db.Put([]byte{0}, []byte{0}))
+		assert.Equal(t, 1, countSegments(t, db))
+		assert.Nil(t, db.Backup(testDBBackupName))
+		db2, err := Open(testDBBackupName, opts)
+		assert.Nil(t, err)
+		v, err := db2.Get([]byte{0})
+		assert.Equal(t, []byte{0}, v)
+		assert.Nil(t, err)
+		assert.Nil(t, db2.Close())
+	})
+
+	run("multiple segments", func(t *testing.T, db *DB) {
+		for i := byte(0); i < 100; i++ {
+			assert.Nil(t, db.Put([]byte{i}, []byte{i}))
+		}
+		assert.Equal(t, 3, countSegments(t, db))
+		assert.Nil(t, db.Backup(testDBBackupName))
+		db2, err := Open(testDBBackupName, opts)
+		assert.Equal(t, 3, countSegments(t, db2))
+		assert.Nil(t, err)
+		for i := byte(0); i < 100; i++ {
+			v, err := db2.Get([]byte{i})
+			assert.Nil(t, err)
+			assert.Equal(t, []byte{i}, v)
+		}
+		assert.Nil(t, db2.Close())
+	})
+}

compaction.go

@@ -1,8 +1,6 @@
 package pogreb
 
 import (
-	"sync/atomic"
-
 	"github.com/akrylysov/pogreb/internal/errors"
 )
 
@@ -135,11 +133,11 @@ func (db *DB) Compact() (CompactionResult, error) {
 	cr := CompactionResult{}
 
 	// Run only a single compaction at a time.
-	if !atomic.CompareAndSwapInt32(&db.compactionRunning, 0, 1) {
+	if !db.maintenanceMu.TryLock() {
 		return cr, errBusy
 	}
 	defer func() {
-		atomic.StoreInt32(&db.compactionRunning, 0)
+		db.maintenanceMu.Unlock()
 	}()
 
 	db.mu.RLock()

compaction_test.go

@@ -3,6 +3,7 @@ package pogreb
 import (
 	"os"
 	"path/filepath"
+	"runtime"
 	"sync"
 	"sync/atomic"
 	"testing"
@@ -224,15 +225,20 @@ func TestCompaction(t *testing.T) {
 		wg := sync.WaitGroup{}
 		wg.Add(1)
 		db.mu.Lock()
+		var goroutineRunning int32
 		go func() {
 			// The compaction is blocked until we unlock the mutex.
 			defer wg.Done()
+			atomic.StoreInt32(&goroutineRunning, 1)
 			_, err := db.Compact()
 			assert.Nil(t, err)
 		}()
-		// Make sure the compaction is running.
+		// Make sure the compaction goroutine is running.
+		// It's unlikely, but still possible the compaction goroutine doesn't reach the maintenance
+		// lock, which makes the test flaky.
+		runtime.Gosched()
 		assert.CompleteWithin(t, time.Minute, func() bool {
-			return atomic.LoadInt32(&db.compactionRunning) == 1
+			return atomic.LoadInt32(&goroutineRunning) == 1
 		})
 		_, err := db.Compact()
 		assert.Equal(t, errBusy, err)

db.go

@@ -30,17 +30,17 @@ const (
 // DB represents the key-value storage.
 // All DB methods are safe for concurrent use by multiple goroutines.
 type DB struct {
-	mu                sync.RWMutex // Allows multiple database readers or a single writer.
-	opts              *Options
-	index             *index
-	datalog           *datalog
-	lock              fs.LockFile // Prevents opening multiple instances of the same database.
-	hashSeed          uint32
-	metrics           *Metrics
-	syncWrites        bool
-	cancelBgWorker    context.CancelFunc
-	closeWg           sync.WaitGroup
-	compactionRunning int32 // Prevents running compactions concurrently.
+	mu             sync.RWMutex // Allows multiple database readers or a single writer.
+	opts           *Options
+	index          *index
+	datalog        *datalog
+	lock           fs.LockFile // Prevents opening multiple instances of the same database.
+	hashSeed       uint32
+	metrics        *Metrics
+	syncWrites     bool
+	cancelBgWorker context.CancelFunc
+	closeWg        sync.WaitGroup
+	maintenanceMu  sync.Mutex // Ensures there only one maintenance task running at a time.
 }
 
 type dbMeta struct {
@@ -52,7 +52,7 @@ type dbMeta struct {
 func Open(path string, opts *Options) (*DB, error) {
 	opts = opts.copyWithDefaults(path)
 
-	if err := os.MkdirAll(path, 0755); err != nil {
+	if err := opts.rootFS.MkdirAll(path, 0755); err != nil {
 		return nil, err
 	}
 

db_test.go

@@ -41,19 +41,12 @@ func TestMain(m *testing.M) {
 			fmt.Printf("DEBUG\tFS=%T\n", fsys)
 			os.Exit(exitCode)
 		}
+		_ = cleanDir(testDBName)
+		_ = cleanDir(testDBBackupName)
 	}
-	_ = cleanDir(testDBName)
 	os.Exit(0)
 }
 
-func touchFile(fsys fs.FileSystem, path string) error {
-	f, err := fsys.OpenFile(path, os.O_CREATE|os.O_TRUNC, os.FileMode(0640))
-	if err != nil {
-		return err
-	}
-	return f.Close()
-}
-
 func appendFile(path string, data []byte) error {
 	f, err := testFS.OpenFile(path, os.O_RDWR, os.FileMode(0640))
 	if err != nil {
@@ -93,7 +86,7 @@ func cleanDir(path string) error {
 		return err
 	}
 	for _, file := range files {
-		_ = testFS.Remove(filepath.Join(testDBName, file.Name()))
+		_ = testFS.Remove(filepath.Join(path, file.Name()))
 	}
 	return nil
 }

file_test.go

@@ -34,6 +34,10 @@ func (fs *errfs) ReadDir(name string) ([]os.DirEntry, error) {
 	return nil, errfileError
 }
 
+func (fs *errfs) MkdirAll(path string, perm os.FileMode) error {
+	return errfileError
+}
+
 type errfile struct{}
 
 var errfileError = errors.New("errfile error")

fs/fs.go

@@ -14,6 +14,7 @@ var (
 )
 
 // File is the interface compatible with os.File.
+// All methods are not thread-safe, except for ReadAt, Slice and Stat.
 type File interface {
 	io.Closer
 	io.Reader
@@ -60,4 +61,7 @@ type FileSystem interface {
 
 	// CreateLockFile creates a lock file.
 	CreateLockFile(name string, perm os.FileMode) (LockFile, bool, error)
+
+	// MkdirAll creates a directory named path.
+	MkdirAll(path string, perm os.FileMode) error
 }