pageserver - fsync image/delta layers

Ensure image and delta layer files are durable.
Also, fsync the parent directory to ensure the directory entries are
durable.

Author: patins

pageserver/src/layered_repository.rs

@@ -1311,6 +1311,8 @@ impl LayeredTimeline {
             last_record_lsn
         );
 
+        let timeline_dir = File::open(self.conf.timeline_path(&self.timelineid, &self.tenantid))?;
+
         // Take the in-memory layer with the oldest WAL record. If it's older
         // than the threshold, write it out to disk as a new image and delta file.
         // Repeat until all remaining in-memory layers are within the threshold.
@@ -1322,6 +1324,8 @@ impl LayeredTimeline {
         // a lot of memory and/or aren't receiving much updates anymore.
         let mut disk_consistent_lsn = last_record_lsn;
 
+        let mut created_historics = false;
+
         while let Some((oldest_layer, oldest_generation)) = layers.peek_oldest_open() {
             let oldest_pending_lsn = oldest_layer.get_oldest_pending_lsn();
 
@@ -1374,6 +1378,11 @@ impl LayeredTimeline {
             drop(layers);
             let new_historics = frozen.write_to_disk(self)?;
             layers = self.layers.lock().unwrap();
+
+            if !new_historics.is_empty() {
+                created_historics = true;
+            }
+
             if let Some(relish_uploader) = &self.relish_uploader {
                 for label_path in new_historics.iter().filter_map(|layer| layer.path()) {
                     relish_uploader.schedule_upload(self.timelineid, label_path);
@@ -1409,6 +1418,14 @@ impl LayeredTimeline {
             layer.unload()?;
         }
 
+        drop(layers);
+
+        if created_historics {
+            // We must fsync the timeline dir to ensure the directory entries for
+            // new layer files are durable
+            timeline_dir.sync_all()?;
+        }
+
         // Save the metadata, with updated 'disk_consistent_lsn', to a
         // file in the timeline dir. After crash, we will restart WAL
         // streaming and processing from that point.

pageserver/src/layered_repository/delta_layer.rs

@@ -487,7 +487,8 @@ impl DeltaLayer {
         let book = chapter.close()?;
 
         // This flushes the underlying 'buf_writer'.
-        book.close()?;
+        let writer = book.close()?;
+        writer.get_ref().sync_all()?;
 
         trace!("saved {}", &path.display());
 

pageserver/src/layered_repository/image_layer.rs

@@ -337,7 +337,8 @@ impl ImageLayer {
         let book = chapter.close()?;
 
         // This flushes the underlying 'buf_writer'.
-        book.close()?;
+        let writer = book.close()?;
+        writer.get_ref().sync_all()?;
 
         trace!("saved {}", &path.display());