Merge pull request #1619 from rqlite/automatic-vacuum

Support automatic VACUUM

CHANGELOG.md

@@ -1,8 +1,12 @@
+## 8.17.0 (unreleased)
+### New features
+- [PR #1619](https://github.com/rqlite/rqlite/pull/1619): Support automatic `VACUUM` of the SQLite database. Fixes [#1609](https://github.com/rqlite/rqlite/issues/1609).
+
 ## 8.16.8 (January 20th 2024)
 ### Implementation changes and bug fixes
 - [PR #1615](https://github.com/rqlite/rqlite/pull/1615): Add extensive WAL checkpoint test at the database level.
 - [PR #1616](https://github.com/rqlite/rqlite/pull/1616): Add time and checksum based change-detection functions to the database level.
-- [PR #1617](https://github.com/rqlite/rqlite/pull/1617): Add VacuumInto to database layer.
+- [PR #1617](https://github.com/rqlite/rqlite/pull/1617): Add `VacuumInto` to database layer.
 - [PR #1621](https://github.com/rqlite/rqlite/pull/1621): Fix a panic in the rqlite shell by not stomping on an "outer" error.
 
 ## 8.16.7 (January 18th 2024)

cmd/rqbench/simple_load.sh

@@ -14,7 +14,7 @@ handle_ctrl_c() {
     exit 1
 }
 
-COUNT=10000
+COUNT=1000000
 
 trap 'handle_ctrl_c' SIGINT
 
@@ -29,4 +29,6 @@ $RQBENCH -p "/db/query" -n 150000000 -m 1000 "SELECT COUNT(*) FROM foo" &
 $RQBENCH -p "/db/query" -n 150000000 -m 1000 "SELECT COUNT(*) FROM bar" &
 $RQBENCH -p "/db/query" -n 150000000 -m 1000 "SELECT * FROM qux LIMIT 10" &
 
+$RQBENCH -p "/db/execute" -m 100 -n $COUNT -a $EXECUTE_HOST 'DELETE FROM qux WHERE id IN (SELECT id FROM qux ORDER BY RANDOM() LIMIT 10)'
+
 wait

cmd/rqlited/flags.go

@@ -135,6 +135,9 @@ type Config struct {
 	// FKConstraints enables SQLite foreign key constraints.
 	FKConstraints bool
 
+	// AutoVacInterval sets the automatic VACUUM interval. Use 0s to disable.
+	AutoVacInterval time.Duration
+
 	// RaftLogLevel sets the minimum logging level for the Raft subsystem.
 	RaftLogLevel string
 
@@ -460,6 +463,7 @@ func ParseFlags(name, desc string, build *BuildInfo) (*Config, error) {
 	flag.StringVar(&config.OnDiskPath, "on-disk-path", "", "Path for SQLite on-disk database file. If not set, use a file in data directory")
 	flag.BoolVar(&config.FKConstraints, "fk", false, "Enable SQLite foreign key constraints")
 	flag.BoolVar(&showVersion, "version", false, "Show version information and exit")
+	flag.DurationVar(&config.AutoVacInterval, "auto-vacuum-int", 0, "Period between automatic VACUUMs. It not set, not enabled")
 	flag.BoolVar(&config.RaftNonVoter, "raft-non-voter", false, "Configure as non-voting node")
 	flag.DurationVar(&config.RaftHeartbeatTimeout, "raft-timeout", time.Second, "Raft heartbeat timeout")
 	flag.DurationVar(&config.RaftElectionTimeout, "raft-election-timeout", time.Second, "Raft election timeout")

cmd/rqlited/main.go

@@ -282,6 +282,7 @@ func createStore(cfg *Config, ln *tcp.Layer) (*store.Store, error) {
 	str.BootstrapExpect = cfg.BootstrapExpect
 	str.ReapTimeout = cfg.RaftReapNodeTimeout
 	str.ReapReadOnlyTimeout = cfg.RaftReapReadOnlyNodeTimeout
+	str.AutoVacInterval = cfg.AutoVacInterval
 
 	if store.IsNewNode(cfg.DataPath) {
 		log.Printf("no preexisting node state detected in %s, node may be bootstrapping", cfg.DataPath)

db/swappable_db.go

@@ -0,0 +1,151 @@
+package db
+
+import (
+	"fmt"
+	"io"
+	"os"
+	"sync"
+
+	command "github.com/rqlite/rqlite/v8/command/proto"
+)
+
+// SwappableDB is a wrapper around DB that allows the underlying database to be swapped out
+// in a thread-safe manner.
+type SwappableDB struct {
+	db   *DB
+	dbMu sync.RWMutex
+}
+
+// OpenSwappable returns a new SwappableDB instance, which opens the database at the given path.
+func OpenSwappable(dbPath string, fkEnabled, wal bool) (*SwappableDB, error) {
+	db, err := Open(dbPath, fkEnabled, wal)
+	if err != nil {
+		return nil, err
+	}
+	return &SwappableDB{db: db}, nil
+}
+
+// Swap swaps the underlying database with that at the given path.
+func (s *SwappableDB) Swap(path string, fkConstraints, walEnabled bool) error {
+	if !IsValidSQLiteFile(path) {
+		return fmt.Errorf("invalid SQLite data")
+	}
+
+	s.dbMu.Lock()
+	defer s.dbMu.Unlock()
+	if err := s.db.Close(); err != nil {
+		return fmt.Errorf("failed to close: %s", err)
+	}
+	if err := RemoveFiles(s.db.Path()); err != nil {
+		return fmt.Errorf("failed to remove files: %s", err)
+	}
+	if err := os.Rename(path, s.db.Path()); err != nil {
+		return fmt.Errorf("failed to rename database: %s", err)
+	}
+
+	var db *DB
+	db, err := Open(s.db.Path(), fkConstraints, walEnabled)
+	if err != nil {
+		return fmt.Errorf("open SQLite file failed: %s", err)
+	}
+	s.db = db
+	return nil
+}
+
+// Close closes the underlying database.
+func (s *SwappableDB) Close() error {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.Close()
+}
+
+// Stats returns the underlying database's stats.
+func (s *SwappableDB) Stats() (map[string]interface{}, error) {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.Stats()
+}
+
+// Request calls Request on the underlying database.
+func (s *SwappableDB) Request(req *command.Request, xTime bool) ([]*command.ExecuteQueryResponse, error) {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.Request(req, xTime)
+}
+
+// Execute calls Execute on the underlying database.
+func (s *SwappableDB) Execute(ex *command.Request, xTime bool) ([]*command.ExecuteResult, error) {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.Execute(ex, xTime)
+}
+
+// Query calls Query on the underlying database.
+func (s *SwappableDB) Query(q *command.Request, xTime bool) ([]*command.QueryRows, error) {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.Query(q, xTime)
+}
+
+// VacuumInto calls VacuumInto on the underlying database.
+func (s *SwappableDB) VacuumInto(path string) error {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.VacuumInto(path)
+}
+
+// Backup calls Backup on the underlying database.
+func (s *SwappableDB) Backup(path string, vacuum bool) error {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.Backup(path, vacuum)
+}
+
+// Serialize calls Serialize on the underlying database.
+func (s *SwappableDB) Serialize() ([]byte, error) {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.Serialize()
+}
+
+// StmtReadOnly calls StmtReadOnly on the underlying database.
+func (s *SwappableDB) StmtReadOnly(sql string) (bool, error) {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.StmtReadOnly(sql)
+}
+
+// Checkpoint calls Checkpoint on the underlying database.
+func (s *SwappableDB) Checkpoint() error {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.Checkpoint()
+}
+
+// Path calls Path on the underlying database.
+func (s *SwappableDB) Path() string {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.Path()
+}
+
+// Dump calls Dump on the underlying database.
+func (s *SwappableDB) Dump(w io.Writer) error {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.Dump(w)
+}
+
+// FKEnabled calls FKEnabled on the underlying database.
+func (s *SwappableDB) FKEnabled() bool {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.FKEnabled()
+}
+
+// WALEnabled calls WALEnabled on the underlying database.
+func (s *SwappableDB) WALEnabled() bool {
+	s.dbMu.RLock()
+	defer s.dbMu.RUnlock()
+	return s.db.WALEnabled()
+}

store/command_processor.go

@@ -4,6 +4,7 @@ import (
 	"fmt"
 	"log"
 	"os"
+	"path/filepath"
 
 	"github.com/rqlite/rqlite/v8/command"
 	"github.com/rqlite/rqlite/v8/command/chunking"
@@ -41,8 +42,7 @@ func NewCommandProcessor(logger *log.Logger, dm *chunking.DechunkerManager) *Com
 }
 
 // Process processes the given command against the given database.
-func (c *CommandProcessor) Process(data []byte, pDB **sql.DB) (*proto.Command, bool, interface{}) {
-	db := *pDB
+func (c *CommandProcessor) Process(data []byte, db *sql.SwappableDB) (*proto.Command, bool, interface{}) {
 	cmd := &proto.Command{}
 	if err := command.Unmarshal(data, cmd); err != nil {
 		panic(fmt.Sprintf("failed to unmarshal cluster command: %s", err.Error()))
@@ -76,20 +76,23 @@ func (c *CommandProcessor) Process(data []byte, pDB **sql.DB) (*proto.Command, b
 			panic(fmt.Sprintf("failed to unmarshal load subcommand: %s", err.Error()))
 		}
 
-		// Swap the underlying database to the new one.
-		if err := db.Close(); err != nil {
-			return cmd, false, &fsmGenericResponse{error: fmt.Errorf("failed to close post-load database: %s", err)}
-		}
-		if err := sql.RemoveFiles(db.Path()); err != nil {
-			return cmd, false, &fsmGenericResponse{error: fmt.Errorf("failed to remove existing database files: %s", err)}
+		// create a scratch file in the same directory as s.db.Path()
+		fd, err := os.CreateTemp(filepath.Dir(db.Path()), "rqlilte-load-")
+		if err != nil {
+			return cmd, false, &fsmGenericResponse{error: fmt.Errorf("failed to create temporary database file: %s", err)}
 		}
-
-		newDB, err := createOnDisk(lr.Data, db.Path(), db.FKEnabled(), db.WALEnabled())
+		defer os.Remove(fd.Name())
+		defer fd.Close()
+		_, err = fd.Write(lr.Data)
 		if err != nil {
-			return cmd, false, &fsmGenericResponse{error: fmt.Errorf("failed to create on-disk database: %s", err)}
+			return cmd, false, &fsmGenericResponse{error: fmt.Errorf("failed to write to temporary database file: %s", err)}
 		}
+		fd.Close()
 
-		*pDB = newDB
+		// Swap the underlying database to the new one.
+		if err := db.Swap(fd.Name(), db.FKEnabled(), db.WALEnabled()); err != nil {
+			return cmd, false, &fsmGenericResponse{error: fmt.Errorf("error swapping databases: %s", err)}
+		}
 		return cmd, true, &fsmGenericResponse{}
 	case proto.Command_COMMAND_TYPE_LOAD_CHUNK:
 		var lcr proto.LoadChunkRequest
@@ -129,25 +132,9 @@ func (c *CommandProcessor) Process(data []byte, pDB **sql.DB) (*proto.Command, b
 					c.logger.Printf("invalid chunked database file - ignoring")
 					return cmd, false, &fsmGenericResponse{error: fmt.Errorf("invalid chunked database file - ignoring")}
 				}
-
-				// Close the underlying database before we overwrite it.
-				if err := db.Close(); err != nil {
-					return cmd, false, &fsmGenericResponse{error: fmt.Errorf("failed to close post-load database: %s", err)}
+				if err := db.Swap(path, db.FKEnabled(), db.WALEnabled()); err != nil {
+					return cmd, false, &fsmGenericResponse{error: fmt.Errorf("error swapping databases: %s", err)}
 				}
-				if err := sql.RemoveFiles(db.Path()); err != nil {
-					return cmd, false, &fsmGenericResponse{error: fmt.Errorf("failed to remove existing database files: %s", err)}
-				}
-
-				if err := os.Rename(path, db.Path()); err != nil {
-					return cmd, false, &fsmGenericResponse{error: fmt.Errorf("failed to rename temporary database file: %s", err)}
-				}
-				newDB, err := sql.Open(db.Path(), db.FKEnabled(), db.WALEnabled())
-				if err != nil {
-					return cmd, false, &fsmGenericResponse{error: fmt.Errorf("failed to open new on-disk database: %s", err)}
-				}
-
-				// Swap the underlying database to the new one.
-				*pDB = newDB
 			}
 		}
 		return cmd, true, &fsmGenericResponse{}

store/state.go

@@ -104,7 +104,7 @@ func RecoverNode(dataDir string, logger *log.Logger, logs raft.LogStore, stable
 	}
 
 	// Now, open the database so we can replay any outstanding Raft log entries.
-	db, err := sql.Open(tmpDBPath, false, true)
+	db, err := sql.OpenSwappable(tmpDBPath, false, true)
 	if err != nil {
 		return fmt.Errorf("failed to open temporary database: %s", err)
 	}
@@ -135,7 +135,7 @@ func RecoverNode(dataDir string, logger *log.Logger, logs raft.LogStore, stable
 			return fmt.Errorf("failed to get log at index %d: %v", index, err)
 		}
 		if entry.Type == raft.LogCommand {
-			cmdProc.Process(entry.Data, &db)
+			cmdProc.Process(entry.Data, db)
 		}
 		lastIndex = entry.Index
 		lastTerm = entry.Term

store/store.go

@@ -6,6 +6,7 @@ package store
 import (
 	"bytes"
 	"compress/gzip"
+	"encoding/binary"
 	"errors"
 	"expvar"
 	"fmt"
@@ -85,6 +86,7 @@ const (
 	restoreScratchPattern      = "rqlite-restore-*"
 	bootScatchPattern          = "rqlite-boot-*"
 	backupScatchPattern        = "rqlite-backup-*"
+	vacuumScatchPattern        = "rqlite-vacuum-*"
 	raftDBPath                 = "raft.db" // Changing this will break backwards compatibility.
 	peersPath                  = "raft/peers.json"
 	peersInfoPath              = "raft/peers.info"
@@ -100,6 +102,8 @@ const (
 	raftLogCacheSize           = 512
 	trailingScale              = 1.25
 	observerChanLen            = 50
+
+	lastVacuumTimeKey = "rqlite_last_vacuum"
 )
 
 const (
@@ -111,6 +115,9 @@ const (
 	numWALSnapshotsFailed     = "num_wal_snapshots_failed"
 	numSnapshotsFull          = "num_snapshots_full"
 	numSnapshotsIncremental   = "num_snapshots_incremental"
+	numAutoVacuums            = "num_auto_vacuums"
+	numAutoVacuumsFailed      = "num_auto_vacuums_failed"
+	autoVacuumDuration        = "auto_vacuum_duration"
 	numBoots                  = "num_boots"
 	numBackups                = "num_backups"
 	numLoads                  = "num_loads"
@@ -159,6 +166,9 @@ func ResetStats() {
 	stats.Add(numWALSnapshotsFailed, 0)
 	stats.Add(numSnapshotsFull, 0)
 	stats.Add(numSnapshotsIncremental, 0)
+	stats.Add(numAutoVacuums, 0)
+	stats.Add(numAutoVacuumsFailed, 0)
+	stats.Add(autoVacuumDuration, 0)
 	stats.Add(numBoots, 0)
 	stats.Add(numBackups, 0)
 	stats.Add(numLoads, 0)
@@ -228,12 +238,12 @@ type Store struct {
 	raft    *raft.Raft // The consensus mechanism.
 	ly      Layer
 	raftTn  *NodeTransport
-	raftID  string    // Node ID.
-	dbConf  *DBConfig // SQLite database config.
-	dbPath  string    // Path to underlying SQLite file.
-	walPath string    // Path to WAL file.
-	dbDir   string    // Path to directory containing SQLite file.
-	db      *sql.DB   // The underlying SQLite store.
+	raftID  string           // Node ID.
+	dbConf  *DBConfig        // SQLite database config.
+	dbPath  string           // Path to underlying SQLite file.
+	walPath string           // Path to WAL file.
+	dbDir   string           // Path to directory containing SQLite file.
+	db      *sql.SwappableDB // The underlying SQLite store.
 
 	dechunkManager *chunking.DechunkerManager
 	cmdProc        *CommandProcessor
@@ -301,6 +311,7 @@ type Store struct {
 	ApplyTimeout             time.Duration
 	RaftLogLevel             string
 	NoFreeListSync           bool
+	AutoVacInterval          time.Duration
 
 	// Node-reaping configuration
 	ReapTimeout         time.Duration
@@ -309,6 +320,7 @@ type Store struct {
 	numTrailingLogs uint64
 
 	// For whitebox testing
+	numAutoVacuums  int
 	numIgnoredJoins int
 	numNoops        int
 	numSnapshotsMu  sync.Mutex
@@ -482,16 +494,19 @@ func (s *Store) Open() (retErr error) {
 	s.logger.Printf("first log index: %d, last log index: %d, last applied index: %d, last command log index: %d:",
 		s.firstIdxOnOpen, s.lastIdxOnOpen, s.lastAppliedIdxOnOpen, s.lastCommandIdxOnOpen)
 
-	s.db, err = createOnDisk(nil, s.dbPath, s.dbConf.FKConstraints, true)
+	s.db, err = createOnDisk(s.dbPath, s.dbConf.FKConstraints, true)
 	if err != nil {
 		return fmt.Errorf("failed to create on-disk database: %s", err)
 	}
-	s.logger.Printf("created on-disk database at open")
 
 	// Clean up any files from aborted operations. This tries to catch the case where scratch files
 	// were created in the Raft directory, not cleaned up, and then the node was restarted with an
 	// explicit SQLite path set.
-	for _, pattern := range []string{restoreScratchPattern, bootScatchPattern, backupScatchPattern} {
+	for _, pattern := range []string{
+		restoreScratchPattern,
+		bootScatchPattern,
+		backupScatchPattern,
+		vacuumScatchPattern} {
 		for _, dir := range []string{s.raftDir, s.dbDir} {
 			files, err := filepath.Glob(filepath.Join(dir, pattern))
 			if err != nil {
@@ -530,6 +545,9 @@ func (s *Store) Open() (retErr error) {
 	// Periodically update the applied index for faster startup.
 	s.appliedIdxUpdateDone = s.updateAppliedIndex()
 
+	if err := s.initLastVacuumTime(); err != nil {
+		return fmt.Errorf("failed to initialize last vacuum time: %s", err.Error())
+	}
 	return nil
 }
 
@@ -726,6 +744,16 @@ func (s *Store) State() ClusterState {
 	}
 }
 
+// LastVacuumTime returns the time of the last automatic VACUUM.
+func (s *Store) LastVacuumTime() (time.Time, error) {
+	vt, err := s.boltStore.Get([]byte(lastVacuumTimeKey))
+	if err != nil {
+		return time.Time{}, fmt.Errorf("failed to get last vacuum time: %s", err)
+	}
+	n := int64(binary.LittleEndian.Uint64(vt))
+	return time.Unix(0, n), nil
+}
+
 // Path returns the path to the store's storage directory.
 func (s *Store) Path() string {
 	return s.raftDir
@@ -991,6 +1019,9 @@ func (s *Store) Stats() (map[string]interface{}, error) {
 		"sqlite3":                dbStatus,
 		"db_conf":                s.dbConf,
 	}
+	if lVac, err := s.LastVacuumTime(); err == nil {
+		status["last_vacuum"] = lVac.String()
+	}
 
 	// Snapshot stats may be in flux if a snapshot is in progress. Only
 	// report them if they are available.
@@ -1321,7 +1352,7 @@ func (s *Store) ReadFrom(r io.Reader) (int64, error) {
 		return 0, ErrNotSingleNode
 	}
 
-	// Write the data to a temporary file..
+	// Write the data to a temporary file.
 	f, err := os.CreateTemp(s.dbDir, bootScatchPattern)
 	if err != nil {
 		return 0, err
@@ -1356,20 +1387,9 @@ func (s *Store) ReadFrom(r io.Reader) (int64, error) {
 	}
 
 	// Swap in new database file.
-	if err := s.db.Close(); err != nil {
-		return n, err
-	}
-	if err := sql.RemoveFiles(s.dbPath); err != nil {
-		return n, err
+	if err := s.db.Swap(f.Name(), s.dbConf.FKConstraints, true); err != nil {
+		return n, fmt.Errorf("error swapping database file: %v", err)
 	}
-	if err := os.Rename(f.Name(), s.dbPath); err != nil {
-		return n, err
-	}
-	db, err := sql.Open(s.dbPath, s.dbConf.FKConstraints, true)
-	if err != nil {
-		return n, err
-	}
-	s.db = db
 
 	// Snapshot, so we load the new database into the Raft system.
 	if err := s.snapshotStore.SetFullNeeded(); err != nil {
@@ -1382,6 +1402,55 @@ func (s *Store) ReadFrom(r io.Reader) (int64, error) {
 	return n, nil
 }
 
+// Vacuum performs a VACUUM operation on the underlying database. It is up
+// to the caller to ensure that no writes are taking place during this call.
+func (s *Store) Vacuum() error {
+	fd, err := os.CreateTemp(s.dbDir, vacuumScatchPattern)
+	if err != nil {
+		return err
+	}
+	if err := fd.Close(); err != nil {
+		return err
+	}
+	defer os.Remove(fd.Name())
+	if err := s.db.VacuumInto(fd.Name()); err != nil {
+		return err
+	}
+
+	// Verify that the VACUUMed database is valid.
+	if !sql.IsValidSQLiteFile(fd.Name()) {
+		return fmt.Errorf("invalid SQLite file post VACUUM")
+	}
+
+	// Swap in new database file.
+	if err := s.db.Swap(fd.Name(), s.dbConf.FKConstraints, true); err != nil {
+		return fmt.Errorf("error swapping database file: %v", err)
+	}
+
+	if err := s.snapshotStore.SetFullNeeded(); err != nil {
+		return err
+	}
+	if err := s.setLastVacuumTime(time.Now()); err != nil {
+		return err
+	}
+	return nil
+}
+
+// Database returns a copy of the underlying database. The caller MUST
+// ensure that no transaction is taking place during this call, or an error may
+// be returned. If leader is true, this operation is performed with a read
+// consistency level equivalent to "weak". Otherwise, no guarantees are made
+// about the read consistency level.
+//
+// http://sqlite.org/howtocorrupt.html states it is safe to do this
+// as long as the database is not written to during the call.
+func (s *Store) Database(leader bool) ([]byte, error) {
+	if leader && s.raft.State() != raft.Leader {
+		return nil, ErrNotLeader
+	}
+	return s.db.Serialize()
+}
+
 // Notify notifies this Store that a node is ready for bootstrapping at the
 // given address. Once the number of known nodes reaches the expected level
 // bootstrapping will be attempted using this Store. "Expected level" includes
@@ -1606,6 +1675,24 @@ func (s *Store) remove(id string) error {
 	return f.Error()
 }
 
+func (s *Store) initLastVacuumTime() error {
+	if _, err := s.LastVacuumTime(); err != nil {
+		return s.setLastVacuumTime(time.Now())
+	}
+	return nil
+}
+
+func (s *Store) setLastVacuumTime(t time.Time) error {
+	buf := bytes.NewBuffer(make([]byte, 0, 8))
+	if err := binary.Write(buf, binary.LittleEndian, t.UnixNano()); err != nil {
+		return fmt.Errorf("failed to encode last vacuum time: %s", err)
+	}
+	if err := s.boltStore.Set([]byte(lastVacuumTimeKey), buf.Bytes()); err != nil {
+		return fmt.Errorf("failed to set last vacuum time: %s", err)
+	}
+	return nil
+}
+
 // raftConfig returns a new Raft config for the store.
 func (s *Store) raftConfig() *raft.Config {
 	config := raft.DefaultConfig()
@@ -1702,7 +1789,7 @@ func (s *Store) fsmApply(l *raft.Log) (e interface{}) {
 		s.logger.Printf("first log applied since node start, log at index %d", l.Index)
 	}
 
-	cmd, mutated, r := s.cmdProc.Process(l.Data, &s.db)
+	cmd, mutated, r := s.cmdProc.Process(l.Data, s.db)
 	if mutated {
 		s.dbAppliedIdxMu.Lock()
 		s.dbAppliedIdx = l.Index
@@ -1722,21 +1809,6 @@ func (s *Store) fsmApply(l *raft.Log) (e interface{}) {
 	return r
 }
 
-// Database returns a copy of the underlying database. The caller MUST
-// ensure that no transaction is taking place during this call, or an error may
-// be returned. If leader is true, this operation is performed with a read
-// consistency level equivalent to "weak". Otherwise, no guarantees are made
-// about the read consistency level.
-//
-// http://sqlite.org/howtocorrupt.html states it is safe to do this
-// as long as the database is not written to during the call.
-func (s *Store) Database(leader bool) ([]byte, error) {
-	if leader && s.raft.State() != raft.Leader {
-		return nil, ErrNotLeader
-	}
-	return s.db.Serialize()
-}
-
 // fsmSnapshot returns a snapshot of the database.
 //
 // The system must ensure that no transaction is taking place during this call.
@@ -1748,6 +1820,9 @@ func (s *Store) Database(leader bool) ([]byte, error) {
 // http://sqlite.org/howtocorrupt.html states it is safe to copy or serialize the
 // database as long as no writes to the database are in progress.
 func (s *Store) fsmSnapshot() (fSnap raft.FSMSnapshot, retErr error) {
+	s.queryTxMu.Lock()
+	defer s.queryTxMu.Unlock()
+
 	if err := s.snapshotCAS.Begin(); err != nil {
 		return nil, err
 	}
@@ -1760,6 +1835,24 @@ func (s *Store) fsmSnapshot() (fSnap raft.FSMSnapshot, retErr error) {
 		}
 	}()
 
+	// Automatic VACUUM needed?
+	if avn, err := s.autoVacNeeded(time.Now()); err != nil {
+		return nil, err
+	} else if avn {
+		vacStart := time.Now()
+		if err := s.Vacuum(); err != nil {
+			stats.Add(numAutoVacuumsFailed, 1)
+			return nil, err
+		}
+		s.logger.Printf("database vacuumed in %s", time.Since(vacStart))
+		stats.Get(autoVacuumDuration).(*expvar.Int).Set(time.Since(vacStart).Milliseconds())
+		stats.Add(numAutoVacuums, 1)
+		s.numAutoVacuums++
+		if err := s.setLastVacuumTime(time.Now()); err != nil {
+			return nil, err
+		}
+	}
+
 	fullNeeded, err := s.snapshotStore.FullNeeded()
 	if err != nil {
 		return nil, err
@@ -1771,9 +1864,6 @@ func (s *Store) fsmSnapshot() (fSnap raft.FSMSnapshot, retErr error) {
 		s.numSnapshots++
 	}()
 
-	s.queryTxMu.Lock()
-	defer s.queryTxMu.Unlock()
-
 	var fsmSnapshot raft.FSMSnapshot
 	if fullNeeded {
 		if err := s.db.Checkpoint(); err != nil {
@@ -1867,27 +1957,9 @@ func (s *Store) fsmRestore(rc io.ReadCloser) (retErr error) {
 		return fmt.Errorf("error creating temporary file for restore operation: %v", err)
 	}
 
-	// Must wipe out all pre-existing state if being asked to do a restore, and put
-	// the new database in place.
-	if !sql.IsValidSQLiteFile(tmpFile.Name()) {
-		return fmt.Errorf("invalid SQLite data")
-	}
-	if err := s.db.Close(); err != nil {
-		return fmt.Errorf("failed to close pre-restore database: %s", err)
-	}
-	if err := sql.RemoveFiles(s.db.Path()); err != nil {
-		return fmt.Errorf("failed to remove pre-restore database files: %s", err)
-	}
-	if err := os.Rename(tmpFile.Name(), s.db.Path()); err != nil {
-		return fmt.Errorf("failed to rename restored database: %s", err)
-	}
-
-	var db *sql.DB
-	db, err = sql.Open(s.dbPath, s.dbConf.FKConstraints, true)
-	if err != nil {
-		return fmt.Errorf("open SQLite file during restore: %s", err)
+	if err := s.db.Swap(tmpFile.Name(), s.dbConf.FKConstraints, true); err != nil {
+		return fmt.Errorf("error swapping database file: %v", err)
 	}
-	s.db = db
 	s.logger.Printf("successfully opened database at %s due to restore", s.db.Path())
 
 	// Take conservative approach and assume that everything has changed, so update
@@ -2142,19 +2214,25 @@ func (s *Store) tryCompress(rq command.Requester) ([]byte, bool, error) {
 	return b, compressed, nil
 }
 
-// createOnDisk opens an on-disk database file at the configured path. If b is
-// non-nil, any preexisting file will first be overwritten with those contents.
-// Otherwise, any preexisting file will be removed before the database is opened.
-func createOnDisk(b []byte, path string, fkConstraints, wal bool) (*sql.DB, error) {
+// autoVacNeeded returns true if an automatic VACUUM is needed.
+func (s *Store) autoVacNeeded(t time.Time) (bool, error) {
+	if s.AutoVacInterval == 0 {
+		return false, nil
+	}
+	lvt, err := s.LastVacuumTime()
+	if err != nil {
+		return false, err
+	}
+	return t.Sub(lvt) > s.AutoVacInterval, nil
+}
+
+// createOnDisk opens an on-disk database file at the configured path. Any
+// preexisting file will be removed before the database is opened.
+func createOnDisk(path string, fkConstraints, wal bool) (*sql.SwappableDB, error) {
 	if err := sql.RemoveFiles(path); err != nil {
 		return nil, err
 	}
-	if b != nil {
-		if err := os.WriteFile(path, b, 0660); err != nil {
-			return nil, err
-		}
-	}
-	return sql.Open(path, fkConstraints, wal)
+	return sql.OpenSwappable(path, fkConstraints, wal)
 }
 
 func copyFromReaderToFile(path string, r io.Reader) (int64, error) {

store/store_test.go

@@ -11,6 +11,7 @@ import (
 	"os"
 	"path/filepath"
 	"strings"
+	"sync"
 	"testing"
 	"time"
 
@@ -51,6 +52,68 @@ func Test_OpenStoreSingleNode(t *testing.T) {
 	}
 }
 
+func Test_OpenStoreSingleNode_LastVacuum(t *testing.T) {
+	s, ln := mustNewStore(t)
+	defer s.Close(true)
+	defer ln.Close()
+
+	now := time.Now()
+	if err := s.Open(); err != nil {
+		t.Fatalf("failed to open single-node store: %s", err.Error())
+	}
+	lv, err := s.LastVacuumTime()
+	if err != nil {
+		t.Fatalf("failed to retrieve last vacuum time: %s", err.Error())
+	}
+	if lv.Before(now) {
+		t.Fatalf("last vacuum time is before now, lv: %s, now: %s", lv, now)
+	}
+}
+
+// Test_OpenStoreSingleNode_VacuumFullNeeded tests that running a VACUUM
+// means that a full snapshot is needed.
+func Test_OpenStoreSingleNode_VacuumFullNeeded(t *testing.T) {
+	s, ln := mustNewStore(t)
+	defer s.Close(true)
+	defer ln.Close()
+
+	if err := s.Open(); err != nil {
+		t.Fatalf("failed to open single-node store: %s", err.Error())
+	}
+	defer s.Close(true)
+	if err := s.Bootstrap(NewServer(s.ID(), s.Addr(), true)); err != nil {
+		t.Fatalf("failed to bootstrap single-node store: %s", err.Error())
+	}
+	if _, err := s.WaitForLeader(10 * time.Second); err != nil {
+		t.Fatalf("Error waiting for leader: %s", err)
+	}
+
+	er := executeRequestFromStrings([]string{
+		`CREATE TABLE foo (id INTEGER NOT NULL PRIMARY KEY, name TEXT)`,
+	}, false, false)
+	_, err := s.Execute(er)
+	if err != nil {
+		t.Fatalf("failed to execute on single node: %s", err.Error())
+	}
+
+	if err := s.Snapshot(0); err != nil {
+		t.Fatalf("failed to snapshot store: %s", err.Error())
+	}
+	if fn, err := s.snapshotStore.FullNeeded(); err != nil {
+		t.Fatalf("failed to determine full snapshot needed: %s", err.Error())
+	} else if fn {
+		t.Fatalf("full snapshot marked as needed")
+	}
+	if err := s.Vacuum(); err != nil {
+		t.Fatalf("failed to vacuum database: %s", err.Error())
+	}
+	if fn, err := s.snapshotStore.FullNeeded(); err != nil {
+		t.Fatalf("failed to determine full snapshot needed: %s", err.Error())
+	} else if !fn {
+		t.Fatalf("full snapshot not marked as needed")
+	}
+}
+
 // Test_SingleNodeSQLitePath ensures that basic functionality works when the SQLite database path
 // is explicitly specificed.
 func Test_SingleNodeOnDiskSQLitePath(t *testing.T) {
@@ -1810,6 +1873,197 @@ func Test_SingleNode_WALTriggeredSnapshot(t *testing.T) {
 			t.Fatalf("wrong snapshot store file: %s", f.Name())
 		}
 	}
+}
+
+// Test_SingleNode_SnapshotWithAutoVac tests that a Store correctly operates
+// when performing both Snapshots and Auto-Vacuums.
+func Test_SingleNode_SnapshotWithAutoVac(t *testing.T) {
+	s, ln := mustNewStore(t)
+	defer ln.Close()
+	s.SnapshotThreshold = 8192 // Ensures Snapshot not triggered during testing.
+	if err := s.Open(); err != nil {
+		t.Fatalf("failed to open single-node store: %s", err.Error())
+	}
+	defer s.Close(true)
+	if err := s.Bootstrap(NewServer(s.ID(), s.Addr(), true)); err != nil {
+		t.Fatalf("failed to bootstrap single-node store: %s", err.Error())
+	}
+	if _, err := s.WaitForLeader(10 * time.Second); err != nil {
+		t.Fatalf("Error waiting for leader: %s", err)
+	}
+
+	// Create a table, and insert some data.
+	er := executeRequestFromString(`CREATE TABLE foo (id INTEGER NOT NULL PRIMARY KEY, name TEXT)`,
+		false, false)
+	_, err := s.Execute(er)
+	if err != nil {
+		t.Fatalf("failed to execute on single node: %s", err.Error())
+	}
+	for i := 0; i < 100; i++ {
+		_, err := s.Execute(executeRequestFromString(`INSERT INTO foo(name) VALUES("fiona")`, false, false))
+		if err != nil {
+			t.Fatalf("failed to execute INSERT on single node: %s", err.Error())
+		}
+	}
+
+	// Force an initial snapshot, shouldn't need a full snapshot afterwards
+	if err := s.Snapshot(0); err != nil {
+		t.Fatalf("failed to snapshot single-node store: %s", err.Error())
+	}
+	if fn, err := s.snapshotStore.FullNeeded(); err != nil {
+		t.Fatalf("failed to check if snapshot store needs a full snapshot: %s", err.Error())
+	} else if fn {
+		t.Fatalf("expected snapshot store to not need a full snapshot")
+	}
+
+	// Enable auto-vacuuming.
+	s.AutoVacInterval = 1 * time.Nanosecond
+	if n, err := s.autoVacNeeded(time.Now()); err != nil {
+		t.Fatalf("failed to check if auto-vacuum is needed: %s", err.Error())
+	} else if !n {
+		t.Fatalf("expected auto-vacuum to be needed")
+	}
+
+	// Force another snapshot, this time a VACUUM should be triggered.
+	if err := s.Snapshot(0); err != nil {
+		t.Fatalf("failed to snapshot single-node store: %s", err.Error())
+	}
+	if exp, got := 1, s.numAutoVacuums; exp != got {
+		t.Fatalf("expected %d auto-vacuums, got %d", exp, got)
+	}
+
+	s.AutoVacInterval = 1 * time.Hour // Effectively disable auto-vacuuming.
+	if n, err := s.autoVacNeeded(time.Now()); err != nil {
+		t.Fatalf("failed to check if auto-vacuum is needed: %s", err.Error())
+	} else if n {
+		t.Fatalf("auto-vacuum should not be needed")
+	}
+
+	// Query the data, make sure it looks good after all this.
+	qr := queryRequestFromString("SELECT COUNT(*) FROM foo", false, true)
+	r, err := s.Query(qr)
+	if err != nil {
+		t.Fatalf("failed to query single node: %s", err.Error())
+	}
+	if exp, got := `[{"columns":["COUNT(*)"],"types":["integer"],"values":[[100]]}]`, asJSON(r); exp != got {
+		t.Fatalf("unexpected results for query\nexp: %s\ngot: %s", exp, got)
+	}
+
+	// Make sure the node works across a restart
+	preAVT, err := s.LastVacuumTime()
+	if err != nil {
+		t.Fatalf("failed to get last vacuum time: %s", err.Error())
+	}
+
+	if err := s.Close(true); err != nil {
+		t.Fatalf("failed to close store: %s", err.Error())
+	}
+	if err := s.Open(); err != nil {
+		t.Fatalf("failed to open store: %s", err.Error())
+	}
+	defer s.Close(true)
+	if _, err := s.WaitForLeader(10 * time.Second); err != nil {
+		t.Fatalf("Error waiting for leader: %s", err)
+	}
+
+	postAVT, err := s.LastVacuumTime()
+	if err != nil {
+		t.Fatalf("failed to get last vacuum time: %s", err.Error())
+	}
+	if preAVT != postAVT {
+		t.Fatalf("expected last vacuum time to be the same across restarts")
+	}
+	if n, err := s.autoVacNeeded(time.Now()); err != nil {
+		t.Fatalf("failed to check if auto-vacuum is needed: %s", err.Error())
+	} else if n {
+		t.Fatalf("auto-vacuum should not be needed")
+	}
+
+	// Query the data again, make sure it still looks good after all this.
+	r, err = s.Query(qr)
+	if err != nil {
+		t.Fatalf("failed to query single node: %s", err.Error())
+	}
+	if exp, got := `[{"columns":["COUNT(*)"],"types":["integer"],"values":[[100]]}]`, asJSON(r); exp != got {
+		t.Fatalf("unexpected results for query\nexp: %s\ngot: %s", exp, got)
+	}
+}
+
+func Test_SingleNode_SnapshotWithAutoVac_Stress(t *testing.T) {
+	s, ln := mustNewStore(t)
+	defer ln.Close()
+	s.SnapshotThreshold = 50
+	s.SnapshotInterval = 100 * time.Millisecond
+	s.AutoVacInterval = 500 * time.Millisecond
+
+	if err := s.Open(); err != nil {
+		t.Fatalf("failed to open single-node store: %s", err.Error())
+	}
+	defer s.Close(true)
+	if err := s.Bootstrap(NewServer(s.ID(), s.Addr(), true)); err != nil {
+		t.Fatalf("failed to bootstrap single-node store: %s", err.Error())
+	}
+	if _, err := s.WaitForLeader(10 * time.Second); err != nil {
+		t.Fatalf("Error waiting for leader: %s", err)
+	}
+
+	// Create a table
+	er := executeRequestFromString(`CREATE TABLE foo (id INTEGER NOT NULL PRIMARY KEY, name TEXT)`,
+		false, false)
+	_, err := s.Execute(er)
+	if err != nil {
+		t.Fatalf("failed to execute on single node: %s", err.Error())
+	}
+
+	// Insert a bunch of data concurrently, putting some load on the Store.
+	var wg sync.WaitGroup
+	wg.Add(5)
+	insertFn := func() {
+		defer wg.Done()
+		for i := 0; i < 500; i++ {
+			_, err := s.Execute(executeRequestFromString(`INSERT INTO foo(name) VALUES("fiona")`, false, false))
+			if err != nil {
+				t.Errorf("failed to execute INSERT on single node: %s", err.Error())
+			}
+		}
+	}
+	for i := 0; i < 5; i++ {
+		go insertFn()
+	}
+	wg.Wait()
+	if s.WaitForAllApplied(5*time.Second) != nil {
+		t.Fatalf("failed to wait for all data to be applied")
+	}
+
+	// Query the data, make sure it looks good after all this.
+	qr := queryRequestFromString("SELECT COUNT(*) FROM foo", false, true)
+	qr.Level = proto.QueryRequest_QUERY_REQUEST_LEVEL_STRONG
+	r, err := s.Query(qr)
+	if err != nil {
+		t.Fatalf("failed to query single node: %s", err.Error())
+	}
+	if exp, got := `[{"columns":["COUNT(*)"],"types":["integer"],"values":[[2500]]}]`, asJSON(r); exp != got {
+		t.Fatalf("unexpected results for query\nexp: %s\ngot: %s", exp, got)
+	}
+
+	// Restart the Store, make sure it still works.
+	if err := s.Close(true); err != nil {
+		t.Fatalf("failed to close store: %s", err.Error())
+	}
+	if err := s.Open(); err != nil {
+		t.Fatalf("failed to open store: %s", err.Error())
+	}
+	defer s.Close(true)
+	if _, err := s.WaitForLeader(10 * time.Second); err != nil {
+		t.Fatalf("Error waiting for leader: %s", err)
+	}
+	r, err = s.Query(qr)
+	if err != nil {
+		t.Fatalf("failed to query single node: %s", err.Error())
+	}
+	if exp, got := `[{"columns":["COUNT(*)"],"types":["integer"],"values":[[2500]]}]`, asJSON(r); exp != got {
+		t.Fatalf("unexpected results for query\nexp: %s\ngot: %s", exp, got)
+	}
 
 }