HBASE-27997 Enhance prefetch executor to record region prefetch infor… (#5339)

Signed-off-by: Wellington Chevreuil <wchevreuil@apache.org>
Reviewew-by: Kota-SH <shanmukhaharipriya@gmail.com>

Author: ragarkar

hbase-protocol-shaded/src/main/protobuf/PrefetchPersistence.proto

@@ -27,5 +27,10 @@ option optimize_for = SPEED;
 
 
 message PrefetchedHfileName {
-  map<string, bool> prefetched_files = 1;
+  map<string, RegionFileSizeMap> prefetched_files = 1;
+}
+
+message RegionFileSizeMap {
+  required string region_name = 1;
+  required uint64 region_prefetch_size = 2;
 }

hbase-server/src/main/java/org/apache/hadoop/hbase/io/hfile/HFilePreadReader.java

@@ -132,13 +132,23 @@ public void run() {
                 LOG.warn("Close prefetch stream reader failed, path: " + path, e);
               }
             }
+            String regionName = getRegionName(path);
             PrefetchExecutor.complete(path);
           }
         }
       });
     }
   }
 
+  /*
+   * Get the region name for the given file path. A HFile is always kept under the <region>/<column
+   * family>/<hfile>. To find the region for a given hFile, just find the name of the grandparent
+   * directory.
+   */
+  private static String getRegionName(Path path) {
+    return path.getParent().getParent().getName();
+  }
+
   private static String getPathOffsetEndStr(final Path path, final long offset, final long end) {
     return "path=" + path.toString() + ", offset=" + offset + ", end=" + end;
   }

hbase-server/src/main/java/org/apache/hadoop/hbase/io/hfile/PrefetchProtoUtils.java

@@ -17,7 +17,9 @@
  */
 package org.apache.hadoop.hbase.io.hfile;
 
+import java.util.HashMap;
 import java.util.Map;
+import org.apache.hadoop.hbase.util.Pair;
 
 import org.apache.hadoop.hbase.shaded.protobuf.generated.PersistentPrefetchProtos;
 
@@ -26,8 +28,26 @@ private PrefetchProtoUtils() {
   }
 
   static PersistentPrefetchProtos.PrefetchedHfileName
-    toPB(Map<String, Boolean> prefetchedHfileNames) {
-    return PersistentPrefetchProtos.PrefetchedHfileName.newBuilder()
-      .putAllPrefetchedFiles(prefetchedHfileNames).build();
+    toPB(Map<String, Pair<String, Long>> prefetchedHfileNames) {
+    Map<String, PersistentPrefetchProtos.RegionFileSizeMap> tmpMap = new HashMap<>();
+    prefetchedHfileNames.forEach((hFileName, regionPrefetchMap) -> {
+      PersistentPrefetchProtos.RegionFileSizeMap tmpRegionFileSize =
+        PersistentPrefetchProtos.RegionFileSizeMap.newBuilder()
+          .setRegionName(regionPrefetchMap.getFirst())
+          .setRegionPrefetchSize(regionPrefetchMap.getSecond()).build();
+      tmpMap.put(hFileName, tmpRegionFileSize);
+    });
+    return PersistentPrefetchProtos.PrefetchedHfileName.newBuilder().putAllPrefetchedFiles(tmpMap)
+      .build();
+  }
+
+  static Map<String, Pair<String, Long>>
+    fromPB(Map<String, PersistentPrefetchProtos.RegionFileSizeMap> prefetchHFileNames) {
+    Map<String, Pair<String, Long>> hFileMap = new HashMap<>();
+    prefetchHFileNames.forEach((hFileName, regionPrefetchMap) -> {
+      hFileMap.put(hFileName,
+        new Pair<>(regionPrefetchMap.getRegionName(), regionPrefetchMap.getRegionPrefetchSize()));
+    });
+    return hFileMap;
   }
 }