diff --git a/api.go b/api.go
index f440df003..66c22c962 100644
--- a/api.go
+++ b/api.go
@@ -68,6 +68,10 @@ func (api *API) Holder() *Holder {
 	return api.holder
 }
 
+func (api *API) logger() logger.Logger {
+	return api.server.logger
+}
+
 // apiOption is a functional option type for pilosa.API
 type apiOption func(*API) error
 
diff --git a/api_directive.go b/api_directive.go
index 9c00a1dea..ed13f6b5e 100644
--- a/api_directive.go
+++ b/api_directive.go
@@ -355,6 +355,10 @@ func (api *API) loadTableKeys(ctx context.Context, idx *Index, tkey dax.TableKey
 	qtid := tkey.QualifiedTableID()
 
 	mgr := api.serverlessStorage.GetTableKeyManager(qtid, partition)
+	if mgr.IsLocked() {
+		api.logger().Debugf("skipping loadTableKeys (already held) %s %d", tkey, partition)
+		return nil
+	}
 
 	// load latest snapshot
 	rc, err := mgr.LoadLatestSnapshot()
@@ -428,6 +432,10 @@ func (api *API) loadFieldKeys(ctx context.Context, tkey dax.TableKey, field dax.
 	qtid := tkey.QualifiedTableID()
 
 	mgr := api.serverlessStorage.GetFieldKeyManager(qtid, field)
+	if mgr.IsLocked() {
+		api.logger().Debugf("skipping loadFieldKeys (already held) %s %s", tkey, field)
+		return nil
+	}
 
 	// load latest snapshot
 	rc, err := mgr.LoadLatestSnapshot()
@@ -512,6 +520,11 @@ func (api *API) loadShard(ctx context.Context, tkey dax.TableKey, shard dax.Shar
 	partition := dax.PartitionNum(disco.ShardToShardPartition(string(tkey), uint64(shard), disco.DefaultPartitionN))
 
 	mgr := api.serverlessStorage.GetShardManager(qtid, partition, shard)
+	if mgr.IsLocked() {
+		api.logger().Debugf("skipping loadShard (already held) %s %d", tkey, shard)
+		return nil
+	}
+
 	rc, err := mgr.LoadLatestSnapshot()
 	if err != nil {
 		return errors.Wrap(err, "reading latest snapshot for shard")
diff --git a/dax/docker-compose.yml b/dax/docker-compose.yml
index f97cd4f56..c6ca7f8f7 100644
--- a/dax/docker-compose.yml
+++ b/dax/docker-compose.yml
@@ -35,6 +35,7 @@ services:
       FEATUREBASE_COMPUTER_RUN: "true"
       FEATUREBASE_COMPUTER_CONFIG_MDS_ADDRESS: "mds:8080/mds"
       FEATUREBASE_COMPUTER_CONFIG_DATA_DIR: /dax-data/computer
+      FEATUREBASE_COMPUTER_CONFIG_VERBOSE: true
       FEATUREBASE_BIND: 0.0.0.0:8080
       FEATUREBASE_VERBOSE: "true"
       FEATUREBASE_STORAGE_METHOD: boltdb
diff --git a/dax/storage/storage.go b/dax/storage/storage.go
index 5ee55ade0..0839e9af4 100644
--- a/dax/storage/storage.go
+++ b/dax/storage/storage.go
@@ -200,6 +200,8 @@ type Manager struct {
 	lastWLPos          int
 
 	locked bool
+
+	// dirty bool // TODO(jaffee): dirty bit so we can skip snapshotting if there's nothing in WL
 }
 
 func (m *Manager) initialize() *Manager {
@@ -209,6 +211,13 @@ func (m *Manager) initialize() *Manager {
 	return m
 }
 
+// IsLocked checks to see if this particular instance of the Manager
+// believes it holds the lock. It does not look at the state of
+// underlying storage to verify the lock.
+func (m *Manager) IsLocked() bool {
+	return m.locked
+}
+
 // LoadLatestSnapshot finds the most recent snapshot for this resource
 // and returns a ReadCloser for that snapshot data. If there is no
 // snapshot for this resource it returns nil, nil.
@@ -217,6 +226,7 @@ func (m *Manager) LoadLatestSnapshot() (data io.ReadCloser, err error) {
 	if err != nil {
 		return nil, errors.Wrap(err, "listing snapshots")
 	}
+	m.log.Debugf("LoadLatestSnapshot %s/%s: list: %v", m.bucket, m.key, snaps)
 	m.lastWLPos = 0
 
 	if len(snaps) == 0 {
@@ -251,6 +261,11 @@ func (m *Manager) LoadWriteLog() (data io.ReadCloser, err error) {
 		return nil, errors.New(errors.ErrUncoded, "LoadWriteLog called in inconsistent state, can't tell what version to load from")
 	}
 	wLogs, err := m.writeLogger.List(m.bucket, m.key)
+	if err != nil {
+		return nil, errors.Wrap(err, "listing write logs")
+	}
+
+	m.log.Debugf("LoadWriteLog %s/%s: list: %v", m.bucket, m.key, wLogs)
 
 	versions := make([]int, 0, len(wLogs))
 	for _, log := range wLogs {
@@ -308,6 +323,7 @@ func (m *Manager) LoadWriteLog() (data io.ReadCloser, err error) {
 // means that quite a lot has happened in between LoadWriteLog and
 // Lock, and we should probably just die and start over.
 func (m *Manager) Lock() error {
+	m.log.Debugf("Lock %s/%s", m.bucket, m.key)
 	// lock is sort of arbitrarily on the write log interface
 	if err := m.writeLogger.Lock(m.bucket, m.key); err != nil {
 		return errors.Wrap(err, "acquiring lock")
@@ -320,6 +336,7 @@ func (m *Manager) Lock() error {
 // haven't properly loaded and gotten a lock for the resource
 // we're writing to.
 func (m *Manager) Append(msg []byte) error {
+	m.log.Debugf("Append %s/%s", m.bucket, m.key)
 	if m.latestWLVersion < 0 {
 		return errors.New(errors.ErrUncoded, "can't call append before loading and locking write log")
 	}
@@ -332,6 +349,7 @@ func (m *Manager) Append(msg []byte) error {
 // WL and any that complete after the snapshot are in the
 // incremented WL.
 func (m *Manager) IncrementWLVersion() error {
+	m.log.Debugf("IncrementWLVersion %s/%s", m.bucket, m.key)
 	m.latestWLVersion++
 	m.lastWLPos = -1
 	m.loadWLsPastVersion = -1
@@ -344,6 +362,7 @@ func (m *Manager) IncrementWLVersion() error {
 // truncate any write logs which are now incorporated into the
 // snapshot.
 func (m *Manager) Snapshot(rc io.ReadCloser) error {
+	m.log.Debugf("Snapshot %s/%s", m.bucket, m.key)
 	// latestWLVersion has already been incremented at this point, so
 	// we write that version minus 1.
 	err := m.snapshotter.Write(m.bucket, m.key, m.latestWLVersion-1, rc)
@@ -358,6 +377,7 @@ func (m *Manager) Snapshot(rc io.ReadCloser) error {
 // of reading from translate stores who we're hoping to off in the
 // next season.
 func (m *Manager) SnapshotTo(wt io.WriterTo) error {
+	m.log.Debugf("SnapshotTo %s/%s", m.bucket, m.key)
 	err := m.snapshotter.WriteTo(m.bucket, m.key, m.latestWLVersion-1, wt)
 	if err != nil {
 		return errors.Wrap(err, "writing snapshot SnapshotTo")
@@ -373,6 +393,7 @@ func (m *Manager) SnapshotTo(wt io.WriterTo) error {
 // should have those removed by the operating system when the
 // process exits anyway.
 func (m *Manager) Unlock() error {
+	m.log.Debugf("Unlock %s/%s", m.bucket, m.key)
 	if !m.locked {
 		return errors.New(errors.ErrUncoded, "resource was not locked")
 	}
diff --git a/server.go b/server.go
index f707bb8ac..bb0ac7d58 100644
--- a/server.go
+++ b/server.go
@@ -806,6 +806,7 @@ func (s *Server) Close() error {
 		var errh, errd error
 		var errhs error
 		var errc error
+		var errSS error
 
 		if s.cluster != nil {
 			errc = s.cluster.close()
@@ -817,7 +818,9 @@ func (s *Server) Close() error {
 		if s.holder != nil {
 			errh = s.holder.Close()
 		}
-		errSS := s.serverlessStorage.RemoveAll()
+		if s.serverlessStorage != nil {
+			errSS = s.serverlessStorage.RemoveAll()
+		}
 
 		// prefer to return holder error over cluster
 		// error. This order is somewhat arbitrary. It would be better if we had