HDDS-1624 : Refactor operations inside the bucket lock in OM key write. (#882)

hadoop-hdds/common/src/main/java/org/apache/hadoop/utils/UniqueId.java

@@ -0,0 +1,69 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ *  with the License.  You may obtain a copy of the License at
+ *
+ *      http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ *
+ */
+
+package org.apache.hadoop.utils;
+
+import org.apache.hadoop.hdds.HddsUtils;
+
+/**
+ * This class uses system current time milliseconds to generate unique id.
+ */
+public final class UniqueId {
+    /*
+     * When we represent time in milliseconds using 'long' data type,
+     * the LSB bits are used. Currently we are only using 44 bits (LSB),
+     * 20 bits (MSB) are not used.
+     * We will exhaust this 44 bits only when we are in year 2525,
+     * until then we can safely use this 20 bits (MSB) for offset to generate
+     * unique id within millisecond.
+     *
+     * Year        : Mon Dec 31 18:49:04 IST 2525
+     * TimeInMillis: 17545641544247
+     * Binary Representation:
+     *   MSB (20 bits): 0000 0000 0000 0000 0000
+     *   LSB (44 bits): 1111 1111 0101 0010 1001 1011 1011 0100 1010 0011 0111
+     *
+     * We have 20 bits to run counter, we should exclude the first bit (MSB)
+     * as we don't want to deal with negative values.
+     * To be on safer side we will use 'short' data type which is of length
+     * 16 bits and will give us 65,536 values for offset.
+     *
+     */
+
+  private static volatile short offset = 0;
+
+  /**
+   * Private constructor so that no one can instantiate this class.
+   */
+  private UniqueId() {}
+
+  /**
+   * Calculate and returns next unique id based on System#currentTimeMillis.
+   *
+   * @return unique long value
+   */
+  public static synchronized long next() {
+    long utcTime = HddsUtils.getUtcTime();
+    if ((utcTime & 0xFFFF000000000000L) == 0) {
+      return utcTime << Short.SIZE | (offset++ & 0x0000FFFF);
+    }
+    throw new RuntimeException("Got invalid UTC time," +
+        " cannot generate unique Id. UTC Time: " + utcTime);
+  }
+}

hadoop-hdds/common/src/main/java/org/apache/hadoop/utils/db/RDBTable.java

@@ -119,6 +119,16 @@ class RDBTable implements Table<byte[], byte[]> {
     }
   }
 
+  @Override
+  public boolean isExist(byte[] key) throws IOException {
+    try {
+      return db.get(handle, key) != null;
+    } catch (RocksDBException e) {
+      throw toIOException(
+          "Error in accessing DB. ", e);
+    }
+  }
+
   @Override
   public byte[] get(byte[] key) throws IOException {
     try {

hadoop-hdds/common/src/main/java/org/apache/hadoop/utils/db/Table.java

@@ -58,6 +58,16 @@ public interface Table<KEY, VALUE> extends AutoCloseable {
    */
   boolean isEmpty() throws IOException;
 
+  /**
+   * Check if a given key exists in Metadata store.
+   * (Optimization to save on data deserialization)
+   * A lock on the key / bucket needs to be acquired before invoking this API.
+   * @param key metadata key
+   * @return true if the metadata store contains a key.
+   * @throws IOException on Failure
+   */
+  boolean isExist(KEY key) throws IOException;
+
   /**
    * Returns the value mapped to the given key in byte array or returns null
    * if the key is not found.

hadoop-hdds/common/src/main/java/org/apache/hadoop/utils/db/TypedTable.java

@@ -79,6 +79,13 @@ public class TypedTable<KEY, VALUE> implements Table<KEY, VALUE> {
     return rawTable.isEmpty();
   }
 
+  @Override
+  public boolean isExist(KEY key) throws IOException {
+    CacheValue<VALUE> cacheValue= cache.get(new CacheKey<>(key));
+    return (cacheValue != null && cacheValue.getValue() != null) ||
+        rawTable.isExist(codecRegistry.asRawData(key));
+  }
+
   /**
    * Returns the value mapped to the given key in byte array or returns null
    * if the key is not found.

hadoop-hdds/common/src/test/java/org/apache/hadoop/utils/db/TestRDBTableStore.java

@@ -51,7 +51,7 @@ public class TestRDBTableStore {
       Arrays.asList(DFSUtil.bytes2String(RocksDB.DEFAULT_COLUMN_FAMILY),
           "First", "Second", "Third",
           "Fourth", "Fifth",
-          "Sixth");
+          "Sixth", "Seventh");
   @Rule
   public TemporaryFolder folder = new TemporaryFolder();
   private RDBStore rdbStore = null;
@@ -228,4 +228,23 @@ public class TestRDBTableStore {
       }
     }
   }
+
+  @Test
+  public void testIsExist() throws Exception {
+    try (Table<byte[], byte[]> testTable = rdbStore.getTable("Seventh")) {
+      byte[] key =
+          RandomStringUtils.random(10).getBytes(StandardCharsets.UTF_8);
+      byte[] value =
+          RandomStringUtils.random(10).getBytes(StandardCharsets.UTF_8);
+      testTable.put(key, value);
+      Assert.assertTrue(testTable.isExist(key));
+
+      testTable.delete(key);
+      Assert.assertFalse(testTable.isExist(key));
+
+      byte[] invalidKey =
+          RandomStringUtils.random(5).getBytes(StandardCharsets.UTF_8);
+      Assert.assertFalse(testTable.isExist(invalidKey));
+    }
+  }
 }

hadoop-hdds/common/src/test/java/org/apache/hadoop/utils/db/TestTypedRDBTableStore.java

@@ -55,7 +55,7 @@ public class TestTypedRDBTableStore {
       Arrays.asList(DFSUtil.bytes2String(RocksDB.DEFAULT_COLUMN_FAMILY),
           "First", "Second", "Third",
           "Fourth", "Fifth",
-          "Sixth", "Seven");
+          "Sixth", "Seven", "Eighth");
   @Rule
   public TemporaryFolder folder = new TemporaryFolder();
   private RDBStore rdbStore = null;
@@ -316,4 +316,39 @@ public class TestTypedRDBTableStore {
 
     }
   }
+
+  @Test
+  public void testIsExist() throws Exception {
+    try (Table<String, String> testTable = createTypedTable(
+        "Eighth")) {
+      String key =
+          RandomStringUtils.random(10);
+      String value = RandomStringUtils.random(10);
+      testTable.put(key, value);
+      Assert.assertTrue(testTable.isExist(key));
+
+      String invalidKey = key + RandomStringUtils.random(1);
+      Assert.assertFalse(testTable.isExist(invalidKey));
+
+      testTable.delete(key);
+      Assert.assertFalse(testTable.isExist(key));
+    }
+  }
+
+  @Test
+  public void testIsExistCache() throws Exception {
+    try (Table<String, String> testTable = createTypedTable(
+        "Eighth")) {
+      String key =
+          RandomStringUtils.random(10);
+      String value = RandomStringUtils.random(10);
+      testTable.addCacheEntry(new CacheKey<>(key),
+          new CacheValue<>(Optional.of(value), 1L));
+      Assert.assertTrue(testTable.isExist(key));
+
+      testTable.addCacheEntry(new CacheKey<>(key),
+          new CacheValue<>(Optional.absent(), 1L));
+      Assert.assertFalse(testTable.isExist(key));
+    }
+  }
 }

hadoop-hdds/server-scm/src/main/java/org/apache/hadoop/hdds/scm/block/BlockManagerImpl.java

@@ -26,7 +26,6 @@ import java.util.concurrent.TimeUnit;
 import javax.management.ObjectName;
 import org.apache.hadoop.conf.Configuration;
 import org.apache.hadoop.conf.StorageUnit;
-import org.apache.hadoop.hdds.HddsUtils;
 import org.apache.hadoop.hdds.client.BlockID;
 import org.apache.hadoop.hdds.client.ContainerBlockID;
 import org.apache.hadoop.hdds.protocol.proto.HddsProtos.ReplicationFactor;
@@ -46,6 +45,7 @@ import org.apache.hadoop.hdds.scm.pipeline.PipelineNotFoundException;
 import org.apache.hadoop.hdds.scm.server.StorageContainerManager;
 import org.apache.hadoop.metrics2.util.MBeans;
 import org.apache.hadoop.util.StringUtils;
+import org.apache.hadoop.utils.UniqueId;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
@@ -354,47 +354,4 @@ public class BlockManagerImpl implements BlockManager, BlockmanagerMXBean {
   /**
    * This class uses system current time milliseconds to generate unique id.
    */
-  public static final class UniqueId {
-    /*
-     * When we represent time in milliseconds using 'long' data type,
-     * the LSB bits are used. Currently we are only using 44 bits (LSB),
-     * 20 bits (MSB) are not used.
-     * We will exhaust this 44 bits only when we are in year 2525,
-     * until then we can safely use this 20 bits (MSB) for offset to generate
-     * unique id within millisecond.
-     *
-     * Year        : Mon Dec 31 18:49:04 IST 2525
-     * TimeInMillis: 17545641544247
-     * Binary Representation:
-     *   MSB (20 bits): 0000 0000 0000 0000 0000
-     *   LSB (44 bits): 1111 1111 0101 0010 1001 1011 1011 0100 1010 0011 0111
-     *
-     * We have 20 bits to run counter, we should exclude the first bit (MSB)
-     * as we don't want to deal with negative values.
-     * To be on safer side we will use 'short' data type which is of length
-     * 16 bits and will give us 65,536 values for offset.
-     *
-     */
-
-    private static volatile short offset = 0;
-
-    /**
-     * Private constructor so that no one can instantiate this class.
-     */
-    private UniqueId() {}
-
-    /**
-     * Calculate and returns next unique id based on System#currentTimeMillis.
-     *
-     * @return unique long value
-     */
-    public static synchronized long next() {
-      long utcTime = HddsUtils.getUtcTime();
-      if ((utcTime & 0xFFFF000000000000L) == 0) {
-        return utcTime << Short.SIZE | (offset++ & 0x0000FFFF);
-      }
-      throw new RuntimeException("Got invalid UTC time," +
-          " cannot generate unique Id. UTC Time: " + utcTime);
-    }
-  }
 }

hadoop-ozone/ozone-manager/src/main/java/org/apache/hadoop/ozone/om/KeyManagerImpl.java

@@ -84,6 +84,7 @@ import org.apache.hadoop.ozone.protocol.proto.OzoneManagerProtocolProtos
     .PartKeyInfo;
 import org.apache.hadoop.util.Time;
 import org.apache.hadoop.utils.BackgroundService;
+import org.apache.hadoop.utils.UniqueId;
 import org.apache.hadoop.utils.db.BatchOperation;
 import org.apache.hadoop.utils.db.DBStore;
 import org.apache.hadoop.utils.db.CodecRegistry;
@@ -406,10 +407,29 @@ public class KeyManagerImpl implements KeyManager {
     String keyName = args.getKeyName();
     validateBucket(volumeName, bucketName);
 
-    long currentTime = Time.monotonicNowNanos();
+    long currentTime = UniqueId.next();
     OmKeyInfo keyInfo;
-    String openKey;
     long openVersion;
+    // NOTE size of a key is not a hard limit on anything, it is a value that
+    // client should expect, in terms of current size of key. If client sets
+    // a value, then this value is used, otherwise, we allocate a single
+    // block which is the current size, if read by the client.
+    final long size = args.getDataSize() >= 0 ?
+        args.getDataSize() : scmBlockSize;
+    final List<OmKeyLocationInfo> locations = new ArrayList<>();
+
+    ReplicationFactor factor = args.getFactor();
+    if (factor == null) {
+      factor = useRatis ? ReplicationFactor.THREE : ReplicationFactor.ONE;
+    }
+
+    ReplicationType type = args.getType();
+    if (type == null) {
+      type = useRatis ? ReplicationType.RATIS : ReplicationType.STAND_ALONE;
+    }
+
+    String dbKeyName = metadataManager.getOzoneKey(
+        args.getVolumeName(), args.getBucketName(), args.getKeyName());
 
     FileEncryptionInfo encInfo;
 
@@ -417,37 +437,7 @@ public class KeyManagerImpl implements KeyManager {
       metadataManager.getLock().acquireBucketLock(volumeName, bucketName);
       OmBucketInfo bucketInfo = getBucketInfo(volumeName, bucketName);
       encInfo = getFileEncryptionInfo(bucketInfo);
-      // NOTE size of a key is not a hard limit on anything, it is a value that
-      // client should expect, in terms of current size of key. If client sets
-      // a value, then this value is used, otherwise, we allocate a single
-      // block which is the current size, if read by the client.
-      long size = args.getDataSize() >= 0 ? args.getDataSize() : scmBlockSize;
-      List<OmKeyLocationInfo> locations = new ArrayList<>();
-      if (args.getIsMultipartKey()) {
-        keyInfo = prepareMultipartKeyInfo(args, size, locations, encInfo);
-        //TODO args.getMetadata
-      } else {
-        keyInfo = prepareKeyInfo(args, size, locations, encInfo);
-      }
-
-      openVersion = keyInfo.getLatestVersionLocations().getVersion();
-      openKey = metadataManager.getOpenKey(
-          volumeName, bucketName, keyName, currentTime);
-      if (metadataManager.getOpenKeyTable().get(openKey) != null) {
-        // This should not happen. If this condition is satisfied, it means
-        // that we have generated a same openKeyId (i.e. currentTime) for two
-        // different client who are trying to write the same key at the same
-        // time. The chance of this happening is very, very minimal.
-
-        // Do we really need this check? Can we avoid this to gain some
-        // minor performance improvement?
-        LOG.warn("Cannot allocate key. The generated open key id is already" +
-            "used for the same key which is currently being written.");
-        throw new OMException("Cannot allocate key. Not able to get a valid" +
-            "open key id.", ResultCodes.KEY_ALLOCATION_ERROR);
-      }
-      LOG.debug("Key {} allocated in volume {} bucket {}",
-          keyName, volumeName, bucketName);
+      keyInfo = prepareKeyInfo(args, dbKeyName, size, locations, encInfo);
     } catch (OMException e) {
       throw e;
     } catch (IOException ex) {
@@ -457,7 +447,14 @@ public class KeyManagerImpl implements KeyManager {
     } finally {
       metadataManager.getLock().releaseBucketLock(volumeName, bucketName);
     }
-
+    if (keyInfo == null) {
+      // the key does not exist, create a new object, the new blocks are the
+      // version 0
+      keyInfo = createKeyInfo(args, locations, factor, type, size, encInfo);
+    }
+    openVersion = keyInfo.getLatestVersionLocations().getVersion();
+    LOG.debug("Key {} allocated in volume {} bucket {}",
+        keyName, volumeName, bucketName);
     allocateBlockInKey(keyInfo, args.getDataSize(), currentTime);
     return new OpenKeySession(currentTime, keyInfo, openVersion);
   }
@@ -485,33 +482,21 @@ public class KeyManagerImpl implements KeyManager {
     }
   }
 
-  private OmKeyInfo prepareKeyInfo(OmKeyArgs args, long size,
+  private OmKeyInfo prepareKeyInfo(
+      OmKeyArgs keyArgs, String dbKeyName, long size,
       List<OmKeyLocationInfo> locations, FileEncryptionInfo encInfo)
       throws IOException {
-    ReplicationFactor factor = args.getFactor();
-    ReplicationType type = args.getType();
-    OmKeyInfo keyInfo;
-    // If user does not specify a replication strategy or
-    // replication factor, OM will use defaults.
-    if (factor == null) {
-      factor = useRatis ? ReplicationFactor.THREE : ReplicationFactor.ONE;
-    }
-    if (type == null) {
-      type = useRatis ? ReplicationType.RATIS : ReplicationType.STAND_ALONE;
-    }
-    String objectKey = metadataManager.getOzoneKey(
-        args.getVolumeName(), args.getBucketName(), args.getKeyName());
-    keyInfo = metadataManager.getKeyTable().get(objectKey);
-    if (keyInfo != null) {
+    OmKeyInfo keyInfo = null;
+    if (keyArgs.getIsMultipartKey()) {
+      keyInfo = prepareMultipartKeyInfo(keyArgs, size, locations, encInfo);
+      //TODO args.getMetadata
+    } else if (metadataManager.getKeyTable().isExist(dbKeyName)) {
+      keyInfo = metadataManager.getKeyTable().get(dbKeyName);
       // the key already exist, the new blocks will be added as new version
       // when locations.size = 0, the new version will have identical blocks
       // as its previous version
       keyInfo.addNewVersion(locations);
       keyInfo.setDataSize(size + keyInfo.getDataSize());
-    } else {
-      // the key does not exist, create a new object, the new blocks are the
-      // version 0
-      keyInfo = createKeyInfo(args, locations, factor, type, size, encInfo);
     }
     return keyInfo;
   }
@@ -618,13 +603,15 @@ public class KeyManagerImpl implements KeyManager {
     String volumeName = args.getVolumeName();
     String bucketName = args.getBucketName();
     String keyName = args.getKeyName();
-    metadataManager.getLock().acquireBucketLock(volumeName, bucketName);
+    List<OmKeyLocationInfo> locationInfoList = args.getLocationInfoList();
+    String objectKey = metadataManager
+        .getOzoneKey(volumeName, bucketName, keyName);
+    String openKey = metadataManager
+        .getOpenKey(volumeName, bucketName, keyName, clientID);
+    Preconditions.checkNotNull(locationInfoList);
     try {
+      metadataManager.getLock().acquireBucketLock(volumeName, bucketName);
       validateBucket(volumeName, bucketName);
-      String openKey = metadataManager.getOpenKey(volumeName, bucketName,
-          keyName, clientID);
-      String objectKey = metadataManager.getOzoneKey(
-          volumeName, bucketName, keyName);
       OmKeyInfo keyInfo = metadataManager.getOpenKeyTable().get(openKey);
       if (keyInfo == null) {
         throw new OMException("Commit a key without corresponding entry " +
@@ -632,8 +619,6 @@ public class KeyManagerImpl implements KeyManager {
       }
       keyInfo.setDataSize(args.getDataSize());
       keyInfo.setModificationTime(Time.now());
-      List<OmKeyLocationInfo> locationInfoList = args.getLocationInfoList();
-      Preconditions.checkNotNull(locationInfoList);
 
       //update the block length for each block
       keyInfo.updateLocationInfoList(locationInfoList);