diff --git a/compact.go b/compact.go
index 2dc939bc1e..35cb36a63c 100644
--- a/compact.go
+++ b/compact.go
@@ -52,7 +52,7 @@ type Compactor interface {
 	Plan(dir string) ([]string, error)
 
 	// Write persists a Block into a directory.
-	Write(dest string, b BlockReader, mint, maxt int64) error
+	Write(dest string, b BlockReader, mint, maxt int64) (ulid.ULID, error)
 
 	// Compact runs compaction against the provided directories. Must
 	// only be called concurrently with results of Plan().
@@ -321,7 +321,7 @@ func (c *LeveledCompactor) Compact(dest string, dirs ...string) (err error) {
 	return c.write(dest, compactBlockMetas(uid, metas...), blocks...)
 }
 
-func (c *LeveledCompactor) Write(dest string, b BlockReader, mint, maxt int64) error {
+func (c *LeveledCompactor) Write(dest string, b BlockReader, mint, maxt int64) (ulid.ULID, error) {
 	entropy := rand.New(rand.NewSource(time.Now().UnixNano()))
 	uid := ulid.MustNew(ulid.Now(), entropy)
 
@@ -333,7 +333,7 @@ func (c *LeveledCompactor) Write(dest string, b BlockReader, mint, maxt int64) e
 	meta.Compaction.Level = 1
 	meta.Compaction.Sources = []ulid.ULID{uid}
 
-	return c.write(dest, meta, b)
+	return uid, c.write(dest, meta, b)
 }
 
 // instrumentedChunkWriter is used for level 1 compactions to record statistics
diff --git a/db.go b/db.go
index f1972f00bc..4e458dea04 100644
--- a/db.go
+++ b/db.go
@@ -344,7 +344,7 @@ func (db *DB) compact() (changes bool, err error) {
 			mint: mint,
 			maxt: maxt,
 		}
-		if err = db.compactor.Write(db.dir, head, mint, maxt); err != nil {
+		if _, err = db.compactor.Write(db.dir, head, mint, maxt); err != nil {
 			return changes, errors.Wrap(err, "persist head block")
 		}
 		changes = true
@@ -615,7 +615,8 @@ func (db *DB) Snapshot(dir string) error {
 			return errors.Wrap(err, "error snapshotting headblock")
 		}
 	}
-	return db.compactor.Write(dir, db.head, db.head.MinTime(), db.head.MaxTime())
+	_, err := db.compactor.Write(dir, db.head, db.head.MinTime(), db.head.MaxTime())
+	return errors.Wrap(err, "snapshot head block")
 }
 
 // Querier returns a new querier over the data partition for the given time range.