HBASE-27794: Tooling for parsing/reading the prefetch files list file

shankot · shankot · commit 6e07f2620547 · 2023-10-24T14:16:10.000-05:00
diff --git a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/Admin.java b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/Admin.java
@@ -2615,4 +2615,9 @@ List<LogEntry> getLogEntries(Set<ServerName> serverNames, String logType, Server
    * Flush master local region
    */
   void flushMasterStore() throws IOException;
+
+  /**
+   * Get the list of cached files
+   */
+  List<String> getCachedFilesList(ServerName serverName) throws IOException;
 }
diff --git a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AdminOverAsyncAdmin.java b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AdminOverAsyncAdmin.java
@@ -1115,4 +1115,9 @@ public List<LogEntry> getLogEntries(Set<ServerName> serverNames, String logType,
   public void flushMasterStore() throws IOException {
     get(admin.flushMasterStore());
   }
+
+  @Override
+  public List<String> getCachedFilesList(ServerName serverName) throws IOException {
+    return get(admin.getCachedFilesList(serverName));
+  }
 }
diff --git a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncAdmin.java b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncAdmin.java
@@ -1837,4 +1837,9 @@ CompletableFuture<List<LogEntry>> getLogEntries(Set<ServerName> serverNames, Str
    * Flush master local region
    */
   CompletableFuture<Void> flushMasterStore();
+
+  /**
+   * Get the list of cached files
+   */
+  CompletableFuture<List<String>> getCachedFilesList(ServerName serverName);
 }
diff --git a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncHBaseAdmin.java b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/AsyncHBaseAdmin.java
@@ -990,4 +990,9 @@ public CompletableFuture<List<LogEntry>> getLogEntries(Set<ServerName> serverNam
   public CompletableFuture<Void> flushMasterStore() {
     return wrap(rawAdmin.flushMasterStore());
   }
+
+  @Override
+  public CompletableFuture<List<String>> getCachedFilesList(ServerName serverName) {
+    return wrap(rawAdmin.getCachedFilesList(serverName));
+  }
 }
diff --git a/hbase-client/src/main/java/org/apache/hadoop/hbase/client/RawAsyncHBaseAdmin.java b/hbase-client/src/main/java/org/apache/hadoop/hbase/client/RawAsyncHBaseAdmin.java
@@ -132,6 +132,8 @@
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.CompactionSwitchResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.FlushRegionRequest;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.FlushRegionResponse;
+import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetCachedFilesListRequest;
+import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetCachedFilesListResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetOnlineRegionRequest;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetOnlineRegionResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetRegionInfoRequest;
@@ -4453,4 +4455,15 @@ Void> call(controller, stub, request.build(),
           (s, c, req, done) -> s.flushMasterStore(c, req, done), resp -> null))
       .call();
   }
+
+  @Override
+  public CompletableFuture<List<String>> getCachedFilesList(ServerName serverName) {
+    GetCachedFilesListRequest.Builder request = GetCachedFilesListRequest.newBuilder();
+    return this.<List<String>> newAdminCaller()
+      .action((controller, stub) -> this.<GetCachedFilesListRequest, GetCachedFilesListResponse,
+        List<String>> adminCall(controller, stub, request.build(),
+          (s, c, req, done) -> s.getCachedFilesList(c, req, done),
+          resp -> resp.getCachedFilesList()))
+      .serverName(serverName).call();
+  }
 }
diff --git a/hbase-client/src/main/java/org/apache/hadoop/hbase/shaded/protobuf/ProtobufUtil.java b/hbase-client/src/main/java/org/apache/hadoop/hbase/shaded/protobuf/ProtobufUtil.java
@@ -153,6 +153,8 @@
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.AdminService;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.ClearSlowLogResponses;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.CloseRegionRequest;
+import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetCachedFilesListRequest;
+import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetCachedFilesListResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetOnlineRegionRequest;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetOnlineRegionResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetRegionInfoRequest;
@@ -1780,6 +1782,21 @@ public static List<org.apache.hadoop.hbase.client.RegionInfo> getOnlineRegions(
     return getRegionInfos(response);
   }
 
+  /**
+   * Get the list of cached files
+   */
+  public static List<String> getCachedFilesList(final RpcController controller,
+    final AdminService.BlockingInterface admin) throws IOException {
+    GetCachedFilesListRequest request = GetCachedFilesListRequest.newBuilder().build();
+    GetCachedFilesListResponse response = null;
+    try {
+      response = admin.getCachedFilesList(controller, request);
+    } catch (ServiceException se) {
+      throw getRemoteException(se);
+    }
+    return new ArrayList<>(response.getCachedFilesList());
+  }
+
   /**
    * Get the list of region info from a GetOnlineRegionResponse
    * @param proto the GetOnlineRegionResponse
diff --git a/hbase-protocol-shaded/src/main/protobuf/server/region/Admin.proto b/hbase-protocol-shaded/src/main/protobuf/server/region/Admin.proto
@@ -283,6 +283,13 @@ message ExecuteProceduresRequest {
 message ExecuteProceduresResponse {
 }
 
+message GetCachedFilesListRequest {
+}
+
+message GetCachedFilesListResponse {
+  repeated string cached_files = 1;
+}
+
 /**
  * Slow/Large log (LogRequest) use-case specific RPC request. This request payload will be
  * converted in bytes and sent to generic RPC API: GetLogEntries
@@ -405,4 +412,7 @@ service AdminService {
   rpc GetLogEntries(LogRequest)
     returns(LogEntry);
 
+  rpc GetCachedFilesList(GetCachedFilesListRequest)
+    returns(GetCachedFilesListResponse);
+
 }
diff --git a/hbase-server/src/main/java/org/apache/hadoop/hbase/io/hfile/bucket/BucketCache.java b/hbase-server/src/main/java/org/apache/hadoop/hbase/io/hfile/bucket/BucketCache.java
@@ -1936,4 +1936,19 @@ public static Optional<BucketCache> getBuckedCacheFromCacheConfig(CacheConfig ca
     return Optional.empty();
   }
 
+  public List<String> getCachedFilesList() throws IOException {
+    List<String> tmp = new ArrayList<>();
+    File persistedFile = new File(persistencePath);
+    try (FileInputStream fis = new FileInputStream(persistedFile)) {
+      if (fis.skip(ProtobufMagic.lengthOfPBMagic()) != ProtobufMagic.lengthOfPBMagic()) {
+        throw new IOException(
+          "Incorrect number of bytes read while checking for protobuf magic number");
+      }
+      BucketCacheProtos.BucketCacheEntry proto =
+        BucketCacheProtos.BucketCacheEntry.parseDelimitedFrom(fis);
+      tmp.addAll(proto.getPrefetchedFilesMap().keySet());
+    }
+    return tmp;
+  }
+
 }
diff --git a/hbase-server/src/main/java/org/apache/hadoop/hbase/master/MasterRpcServices.java b/hbase-server/src/main/java/org/apache/hadoop/hbase/master/MasterRpcServices.java
@@ -164,6 +164,8 @@
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.ExecuteProceduresResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.FlushRegionRequest;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.FlushRegionResponse;
+import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetCachedFilesListRequest;
+import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetCachedFilesListResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetOnlineRegionRequest;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetOnlineRegionResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetRegionInfoRequest;
@@ -3563,6 +3565,11 @@ public ExecuteProceduresResponse executeProcedures(RpcController controller,
     throw new ServiceException(new DoNotRetryIOException("Unsupported method on master"));
   }
 
+  public GetCachedFilesListResponse getCachedFilesList(RpcController controller,
+    GetCachedFilesListRequest request) throws ServiceException {
+    throw new ServiceException(new DoNotRetryIOException("Unsupported method on master"));
+  }
+
   @Override
   public GetLiveRegionServersResponse getLiveRegionServers(RpcController controller,
     GetLiveRegionServersRequest request) throws ServiceException {
diff --git a/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/HRegionServer.java b/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/HRegionServer.java
@@ -108,7 +108,9 @@
 import org.apache.hadoop.hbase.http.InfoServer;
 import org.apache.hadoop.hbase.io.hfile.BlockCache;
 import org.apache.hadoop.hbase.io.hfile.BlockCacheFactory;
+import org.apache.hadoop.hbase.io.hfile.CombinedBlockCache;
 import org.apache.hadoop.hbase.io.hfile.HFile;
+import org.apache.hadoop.hbase.io.hfile.bucket.BucketCache;
 import org.apache.hadoop.hbase.io.util.MemorySizeUtil;
 import org.apache.hadoop.hbase.ipc.CoprocessorRpcUtils;
 import org.apache.hadoop.hbase.ipc.RpcClient;
@@ -3324,6 +3326,21 @@ public Optional<BlockCache> getBlockCache() {
     return Optional.ofNullable(this.blockCache);
   }
 
+  public Optional<BucketCache> getBucketCacheIfPresent() {
+    if (getBlockCache().isPresent()) {
+      BlockCache bc = getBlockCache().get();
+      if (bc instanceof CombinedBlockCache) {
+        BlockCache l2 = ((CombinedBlockCache) bc).getSecondLevelCache();
+        if (l2 instanceof BucketCache) {
+          return Optional.of((BucketCache) l2);
+        }
+      } else if (bc instanceof BucketCache) {
+        return Optional.of((BucketCache) bc);
+      }
+    }
+    return Optional.empty();
+  }
+
   /**
    * May be null if this is a master which not carry table.
    * @return The cache for mob files used by the regionserver.
diff --git a/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/RSRpcServices.java b/hbase-server/src/main/java/org/apache/hadoop/hbase/regionserver/RSRpcServices.java
@@ -169,6 +169,8 @@
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.ExecuteProceduresResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.FlushRegionRequest;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.FlushRegionResponse;
+import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetCachedFilesListRequest;
+import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetCachedFilesListResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetOnlineRegionRequest;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetOnlineRegionResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetRegionInfoRequest;
@@ -3933,4 +3935,23 @@ public void onConfigurationChange(Configuration conf) {
     super.onConfigurationChange(conf);
     setReloadableGuardrails(conf);
   }
+
+  @Override
+  public GetCachedFilesListResponse getCachedFilesList(RpcController controller,
+    GetCachedFilesListRequest request) throws ServiceException {
+    GetCachedFilesListResponse.Builder responseBuilder = GetCachedFilesListResponse.newBuilder();
+
+    List<String> fullyCachedFiles = new ArrayList<>();
+    server.getBucketCacheIfPresent().ifPresent(bucketCache -> {
+      if (bucketCache.isCachePersistenceEnabled()) {
+        LOG.info("Fetching fully cached files");
+        try {
+          fullyCachedFiles.addAll(bucketCache.getCachedFilesList());
+        } catch (IOException e) {
+          LOG.info("Exception occurred reading the cache persistence file: {}", e.getMessage());
+        }
+      }
+    });
+    return responseBuilder.addAllCachedFiles(fullyCachedFiles).build();
+  }
 }
diff --git a/hbase-server/src/test/java/org/apache/hadoop/hbase/io/hfile/TestPrefetchRSClose.java b/hbase-server/src/test/java/org/apache/hadoop/hbase/io/hfile/TestPrefetchRSClose.java
@@ -23,18 +23,23 @@
 import static org.junit.Assert.assertTrue;
 
 import java.io.File;
+import java.io.FileInputStream;
+import java.util.ArrayList;
 import org.apache.hadoop.conf.Configuration;
 import org.apache.hadoop.fs.Path;
 import org.apache.hadoop.hbase.HBaseClassTestRule;
 import org.apache.hadoop.hbase.HBaseTestingUtil;
 import org.apache.hadoop.hbase.SingleProcessHBaseCluster;
 import org.apache.hadoop.hbase.StartTestingClusterOption;
 import org.apache.hadoop.hbase.TableName;
+import org.apache.hadoop.hbase.client.Admin;
 import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;
 import org.apache.hadoop.hbase.client.Put;
 import org.apache.hadoop.hbase.client.Table;
 import org.apache.hadoop.hbase.client.TableDescriptor;
 import org.apache.hadoop.hbase.client.TableDescriptorBuilder;
+import org.apache.hadoop.hbase.protobuf.ProtobufMagic;
+import org.apache.hadoop.hbase.regionserver.HRegionServer;
 import org.apache.hadoop.hbase.testclassification.IOTests;
 import org.apache.hadoop.hbase.testclassification.LargeTests;
 import org.apache.hadoop.hbase.util.Bytes;
@@ -47,6 +52,8 @@
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
+import org.apache.hadoop.hbase.shaded.protobuf.generated.BucketCacheProtos;
+
 @Category({ IOTests.class, LargeTests.class })
 public class TestPrefetchRSClose {
 
@@ -112,7 +119,20 @@ public void testPrefetchPersistence() throws Exception {
 
     // Default interval for cache persistence is 1000ms. So after 1000ms, both the persistence files
     // should exist.
-    assertTrue(new File(testDir + "/bucket.persistence").exists());
+
+    HRegionServer regionServingRS = cluster.getRegionServer(1).getRegions(tableName).size() == 1
+      ? cluster.getRegionServer(1)
+      : cluster.getRegionServer(0);
+
+    Admin admin = TEST_UTIL.getAdmin();
+    File persistenceFile = new File(testDir + "/bucket.persistence");
+    assertTrue(persistenceFile.exists());
+    FileInputStream fis = new FileInputStream(persistenceFile);
+    fis.skip(ProtobufMagic.lengthOfPBMagic());
+    BucketCacheProtos.BucketCacheEntry proto =
+      BucketCacheProtos.BucketCacheEntry.parseDelimitedFrom(fis);
+    assertEquals(new ArrayList<>(proto.getPrefetchedFilesMap().keySet()),
+      admin.getCachedFilesList(regionServingRS.getServerName()));
 
     // Stop the RS
     cluster.stopRegionServer(0);
diff --git a/hbase-server/src/test/java/org/apache/hadoop/hbase/master/MockRegionServer.java b/hbase-server/src/test/java/org/apache/hadoop/hbase/master/MockRegionServer.java
@@ -97,6 +97,8 @@
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.ExecuteProceduresResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.FlushRegionRequest;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.FlushRegionResponse;
+import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetCachedFilesListRequest;
+import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetCachedFilesListResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetOnlineRegionRequest;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetOnlineRegionResponse;
 import org.apache.hadoop.hbase.shaded.protobuf.generated.AdminProtos.GetRegionInfoRequest;
@@ -452,6 +454,12 @@ public GetOnlineRegionResponse getOnlineRegion(RpcController controller,
     return null;
   }
 
+  @Override
+  public GetCachedFilesListResponse getCachedFilesList(RpcController controller,
+    GetCachedFilesListRequest request) throws ServiceException {
+    return null;
+  }
+
   @Override
   public List<HRegion> getRegions() {
     return null;
diff --git a/hbase-server/src/test/java/org/apache/hadoop/hbase/rsgroup/VerifyingRSGroupAdmin.java b/hbase-server/src/test/java/org/apache/hadoop/hbase/rsgroup/VerifyingRSGroupAdmin.java
@@ -959,6 +959,11 @@ public void flushMasterStore() throws IOException {
     admin.flushMasterStore();
   }
 
+  @Override
+  public List<String> getCachedFilesList(ServerName serverName) throws IOException {
+    return admin.getCachedFilesList(serverName);
+  }
+
   @Override
   public boolean replicationPeerModificationSwitch(boolean on, boolean drainProcedures)
     throws IOException {
diff --git a/hbase-thrift/src/main/java/org/apache/hadoop/hbase/thrift2/client/ThriftAdmin.java b/hbase-thrift/src/main/java/org/apache/hadoop/hbase/thrift2/client/ThriftAdmin.java
@@ -1335,6 +1335,11 @@ public void flushMasterStore() throws IOException {
     throw new NotImplementedException("flushMasterStore not supported in ThriftAdmin");
   }
 
+  @Override
+  public List<String> getCachedFilesList(ServerName serverName) throws IOException {
+    throw new NotImplementedException("getCachedFilesList not supported in ThriftAdmin");
+  }
+
   @Override
   public boolean replicationPeerModificationSwitch(boolean on, boolean drainProcedures)
     throws IOException {

Original file line number	Diff line number	Diff line change
`@@ -1115,4 +1115,9 @@ public List<LogEntry> getLogEntries(Set<ServerName> serverNames, String logType,`
`1115`	`1115`	`public void flushMasterStore() throws IOException {`
`1116`	`1116`	`get(admin.flushMasterStore());`
`1117`	`1117`	`}`
	`1118`	`+`
	`1119`	`+ @Override`
	`1120`	`+ public List<String> getCachedFilesList(ServerName serverName) throws IOException {`
	`1121`	`+ return get(admin.getCachedFilesList(serverName));`
	`1122`	`+ }`
`1118`	`1123`	`}`
Original file line number	Diff line number	Diff line change
`@@ -990,4 +990,9 @@ public CompletableFuture<List<LogEntry>> getLogEntries(Set<ServerName> serverNam`
`990`	`990`	`public CompletableFuture<Void> flushMasterStore() {`
`991`	`991`	`return wrap(rawAdmin.flushMasterStore());`
`992`	`992`	`}`
	`993`	`+`
	`994`	`+ @Override`
	`995`	`+ public CompletableFuture<List<String>> getCachedFilesList(ServerName serverName) {`
	`996`	`+ return wrap(rawAdmin.getCachedFilesList(serverName));`
	`997`	`+ }`
`993`	`998`	`}`