HDDS-6541. [Merge rocksdb in datanode] Per-disk DB location management.

hadoop-hdds/common/src/main/java/org/apache/hadoop/hdds/upgrade/HDDSLayoutFeature.java

@@ -31,7 +31,9 @@ public enum HDDSLayoutFeature implements LayoutFeature {
   INITIAL_VERSION(0, "Initial Layout Version"),
   DATANODE_SCHEMA_V2(1, "Datanode RocksDB Schema Version 2 (with column " +
       "families)"),
-  SCM_HA(2, "Storage Container Manager HA");
+  SCM_HA(2, "Storage Container Manager HA"),
+  DATANODE_SCHEMA_V3(3, "Datanode RocksDB Schema Version 3 (one rocksdb " +
+      "per disk)");
 
   //////////////////////////////  //////////////////////////////
 

hadoop-hdds/common/src/main/java/org/apache/hadoop/ozone/OzoneConfigKeys.java

@@ -328,6 +328,11 @@ public final class OzoneConfigKeys {
   public static final String
       HDDS_DATANODE_METADATA_ROCKSDB_CACHE_SIZE_DEFAULT = "1GB";
 
+  // Specifying the dedicated volumes for per-disk db instances.
+  // For container schema v3 only.
+  public static final String HDDS_DATANODE_CONTAINER_DB_DIR =
+      "hdds.datanode.container.db.dir";
+
   public static final String OZONE_SECURITY_ENABLED_KEY =
       "ozone.security.enabled";
   public static final boolean OZONE_SECURITY_ENABLED_DEFAULT = false;

hadoop-hdds/common/src/main/java/org/apache/hadoop/ozone/OzoneConsts.java

@@ -135,6 +135,7 @@ public final class OzoneConsts {
   public static final String SCM_DB_NAME = "scm.db";
   public static final String OM_DB_BACKUP_PREFIX = "om.db.backup.";
   public static final String SCM_DB_BACKUP_PREFIX = "scm.db.backup.";
+  public static final String CONTAINER_DB_NAME = "container.db";
 
   public static final String STORAGE_DIR_CHUNKS = "chunks";
   public static final String OZONE_DB_CHECKPOINT_REQUEST_FLUSH =

hadoop-hdds/common/src/main/resources/ozone-default.xml

@@ -144,6 +144,20 @@
       tagged explicitly.
     </description>
   </property>
+  <property>
+    <name>hdds.datanode.container.db.dir</name>
+    <value/>
+    <tag>OZONE, CONTAINER, STORAGE, MANAGEMENT</tag>
+    <description>Determines where the per-disk rocksdb instances will be
+      stored. This setting is optional. If unspecified, then rocksdb instances
+      are stored on the same disk as HDDS data.
+      The directories should be tagged with corresponding storage types
+      ([SSD]/[DISK]/[ARCHIVE]/[RAM_DISK]) for storage policies. The default
+      storage type will be DISK if the directory does not have a storage type
+      tagged explicitly. Ideally, this should be mapped to a fast disk
+      like an SSD.
+    </description>
+  </property>
   <property>
     <name>hdds.datanode.dir.du.reserved</name>
     <value/>

hadoop-hdds/container-service/src/main/java/org/apache/hadoop/ozone/HddsDatanodeService.java

@@ -57,7 +57,7 @@
 import org.apache.hadoop.ozone.container.common.DatanodeLayoutStorage;
 import org.apache.hadoop.ozone.container.common.helpers.ContainerUtils;
 import org.apache.hadoop.ozone.container.common.statemachine.DatanodeStateMachine;
-import org.apache.hadoop.ozone.container.common.utils.HddsVolumeUtil;
+import org.apache.hadoop.ozone.container.common.utils.StorageVolumeUtil;
 import org.apache.hadoop.ozone.container.common.volume.HddsVolume;
 import org.apache.hadoop.ozone.container.common.volume.MutableVolumeSet;
 import org.apache.hadoop.ozone.container.common.volume.StorageVolume;
@@ -319,8 +319,8 @@ private void startRatisForTest() throws IOException {
 
     for (Map.Entry<String, StorageVolume> entry : volumeMap.entrySet()) {
       HddsVolume hddsVolume = (HddsVolume) entry.getValue();
-      boolean result = HddsVolumeUtil.checkVolume(hddsVolume, clusterId,
-          clusterId, conf, LOG);
+      boolean result = StorageVolumeUtil.checkVolume(hddsVolume, clusterId,
+          clusterId, conf, LOG, null);
       if (!result) {
         volumeSet.failVolume(hddsVolume.getHddsRootDir().getPath());
       }

hadoop-hdds/container-service/src/main/java/org/apache/hadoop/ozone/container/common/statemachine/DatanodeConfiguration.java

@@ -45,13 +45,17 @@ public class DatanodeConfiguration {
       "hdds.datanode.failed.data.volumes.tolerated";
   public static final String FAILED_METADATA_VOLUMES_TOLERATED_KEY =
       "hdds.datanode.failed.metadata.volumes.tolerated";
+  public static final String FAILED_DB_VOLUMES_TOLERATED_KEY =
+      "hdds.datanode.failed.db.volumes.tolerated";
   public static final String DISK_CHECK_MIN_GAP_KEY =
       "hdds.datanode.disk.check.min.gap";
   public static final String DISK_CHECK_TIMEOUT_KEY =
       "hdds.datanode.disk.check.timeout";
 
   public static final String WAIT_ON_ALL_FOLLOWERS =
       "hdds.datanode.wait.on.all.followers";
+  public static final String CONTAINER_SCHEMA_V3_ENABLED =
+      "hdds.datanode.container.schema.v3.enabled";
 
   static final boolean CHUNK_DATA_VALIDATION_CHECK_DEFAULT = false;
 
@@ -67,6 +71,8 @@ public class DatanodeConfiguration {
   static final long DISK_CHECK_TIMEOUT_DEFAULT =
       Duration.ofMinutes(10).toMillis();
 
+  static final boolean CONTAINER_SCHEMA_V3_ENABLED_DEFAULT = false;
+
   /**
    * Number of threads per volume that Datanode will use for chunk read.
    */
@@ -195,6 +201,17 @@ public void setBlockDeletionLimit(int limit) {
   )
   private int failedMetadataVolumesTolerated = FAILED_VOLUMES_TOLERATED_DEFAULT;
 
+  @Config(key = "failed.db.volumes.tolerated",
+      defaultValue = "-1",
+      type = ConfigType.INT,
+      tags = { DATANODE },
+      description = "The number of db volumes that are allowed to fail "
+          + "before a datanode stops offering service. "
+          + "Config this to -1 means unlimited, but we should have "
+          + "at least one good volume left."
+  )
+  private int failedDbVolumesTolerated = FAILED_VOLUMES_TOLERATED_DEFAULT;
+
   @Config(key = "disk.check.min.gap",
       defaultValue = "15m",
       type = ConfigType.TIME,
@@ -245,6 +262,15 @@ public void setWaitOnAllFollowers(boolean val) {
     this.waitOnAllFollowers = val;
   }
 
+  @Config(key = "container.schema.v3.enabled",
+      defaultValue = "false",
+      type = ConfigType.BOOLEAN,
+      tags = { DATANODE },
+      description = "Enable use of container schema v3(one rocksdb per disk)."
+  )
+  private boolean containerSchemaV3Enabled =
+      CONTAINER_SCHEMA_V3_ENABLED_DEFAULT;
+
   @PostConstruct
   public void validate() {
     if (containerDeleteThreads < 1) {
@@ -277,6 +303,13 @@ public void validate() {
       failedMetadataVolumesTolerated = FAILED_VOLUMES_TOLERATED_DEFAULT;
     }
 
+    if (failedDbVolumesTolerated < -1) {
+      LOG.warn(FAILED_DB_VOLUMES_TOLERATED_KEY +
+              "must be greater than -1 and was set to {}. Defaulting to {}",
+          failedDbVolumesTolerated, FAILED_VOLUMES_TOLERATED_DEFAULT);
+      failedDbVolumesTolerated = FAILED_VOLUMES_TOLERATED_DEFAULT;
+    }
+
     if (diskCheckMinGap < 0) {
       LOG.warn(DISK_CHECK_MIN_GAP_KEY +
               " must be greater than zero and was set to {}. Defaulting to {}",
@@ -325,6 +358,14 @@ public void setFailedMetadataVolumesTolerated(int failedVolumesTolerated) {
     this.failedMetadataVolumesTolerated = failedVolumesTolerated;
   }
 
+  public int getFailedDbVolumesTolerated() {
+    return failedDbVolumesTolerated;
+  }
+
+  public void setFailedDbVolumesTolerated(int failedVolumesTolerated) {
+    this.failedDbVolumesTolerated = failedVolumesTolerated;
+  }
+
   public Duration getDiskCheckMinGap() {
     return Duration.ofMillis(diskCheckMinGap);
   }
@@ -372,4 +413,12 @@ public void setNumReadThreadPerVolume(int threads) {
   public int getNumReadThreadPerVolume() {
     return numReadThreadPerVolume;
   }
+
+  public boolean getContainerSchemaV3Enabled() {
+    return this.containerSchemaV3Enabled;
+  }
+
+  public void setContainerSchemaV3Enabled(boolean containerSchemaV3Enabled) {
+    this.containerSchemaV3Enabled = containerSchemaV3Enabled;
+  }
 }

hadoop-hdds/container-service/src/main/java/org/apache/hadoop/ozone/container/common/statemachine/DatanodeStateMachine.java

@@ -316,8 +316,8 @@ private void startStateMachineThread() throws IOException {
   public void handleFatalVolumeFailures() {
     LOG.error("DatanodeStateMachine Shutdown due to too many bad volumes, "
         + "check " + DatanodeConfiguration.FAILED_DATA_VOLUMES_TOLERATED_KEY
-        + " and "
-        + DatanodeConfiguration.FAILED_METADATA_VOLUMES_TOLERATED_KEY);
+        + " and " + DatanodeConfiguration.FAILED_METADATA_VOLUMES_TOLERATED_KEY
+        + " and " + DatanodeConfiguration.FAILED_DB_VOLUMES_TOLERATED_KEY);
     hddsDatanodeStopService.stopService();
   }
 

hadoop-hdds/container-service/src/main/java/org/apache/hadoop/ozone/container/common/states/endpoint/VersionEndpointTask.java

@@ -17,18 +17,17 @@
 package org.apache.hadoop.ozone.container.common.states.endpoint;
 
 import java.io.IOException;
-import java.util.Map;
 import java.util.concurrent.Callable;
 
 import org.apache.hadoop.hdds.conf.ConfigurationSource;
 import org.apache.hadoop.hdds.protocol.proto.StorageContainerDatanodeProtocolProtos.SCMVersionResponseProto;
 import org.apache.hadoop.ozone.OzoneConsts;
 import org.apache.hadoop.ozone.container.common.statemachine.EndpointStateMachine;
-import org.apache.hadoop.ozone.container.common.utils.HddsVolumeUtil;
-import org.apache.hadoop.ozone.container.common.volume.HddsVolume;
+import org.apache.hadoop.ozone.container.common.utils.StorageVolumeUtil;
 import org.apache.hadoop.ozone.container.common.volume.MutableVolumeSet;
 import org.apache.hadoop.ozone.container.common.volume.StorageVolume;
 import org.apache.hadoop.ozone.container.ozoneimpl.OzoneContainer;
+import org.apache.hadoop.ozone.container.upgrade.VersionedDatanodeFeatures.SchemaV3;
 import org.apache.hadoop.ozone.protocol.VersionResponse;
 import org.apache.hadoop.util.DiskChecker.DiskOutOfSpaceException;
 
@@ -78,36 +77,17 @@ public EndpointStateMachine.EndPointStates call() throws Exception {
           String scmId = response.getValue(OzoneConsts.SCM_ID);
           String clusterId = response.getValue(OzoneConsts.CLUSTER_ID);
 
-          // Check volumes
-          MutableVolumeSet volumeSet = ozoneContainer.getVolumeSet();
-          volumeSet.writeLock();
-          try {
-            Map<String, StorageVolume> volumeMap = volumeSet.getVolumeMap();
+          Preconditions.checkNotNull(scmId,
+              "Reply from SCM: scmId cannot be null");
+          Preconditions.checkNotNull(clusterId,
+              "Reply from SCM: clusterId cannot be null");
 
-            Preconditions.checkNotNull(scmId,
-                "Reply from SCM: scmId cannot be null");
-            Preconditions.checkNotNull(clusterId,
-                "Reply from SCM: clusterId cannot be null");
-
-            // If version file does not exist
-            // create version file and also set scm ID or cluster ID.
-            for (Map.Entry<String, StorageVolume> entry
-                : volumeMap.entrySet()) {
-              StorageVolume volume = entry.getValue();
-              boolean result = HddsVolumeUtil.checkVolume((HddsVolume) volume,
-                  scmId, clusterId, configuration, LOG);
-              if (!result) {
-                volumeSet.failVolume(volume.getStorageDir().getPath());
-              }
-            }
-            if (volumeSet.getVolumesList().size() == 0) {
-              // All volumes are in inconsistent state
-              throw new DiskOutOfSpaceException(
-                  "All configured Volumes are in Inconsistent State");
-            }
-          } finally {
-            volumeSet.writeUnlock();
+          // Check DbVolumes
+          if (SchemaV3.isFinalizedAndEnabled(configuration)) {
+            checkVolumeSet(ozoneContainer.getDbVolumeSet(), scmId, clusterId);
           }
+          // Check HddsVolumes
+          checkVolumeSet(ozoneContainer.getVolumeSet(), scmId, clusterId);
 
           // Start the container services after getting the version information
           ozoneContainer.start(clusterId);
@@ -129,4 +109,32 @@ public EndpointStateMachine.EndPointStates call() throws Exception {
     }
     return rpcEndPoint.getState();
   }
+
+  private void checkVolumeSet(MutableVolumeSet volumeSet,
+      String scmId, String clusterId) throws DiskOutOfSpaceException {
+    if (volumeSet == null) {
+      return;
+    }
+
+    volumeSet.writeLock();
+    try {
+      // If version file does not exist
+      // create version file and also set scm ID or cluster ID.
+      for (StorageVolume volume : volumeSet.getVolumeMap().values()) {
+        boolean result = StorageVolumeUtil.checkVolume(volume,
+            scmId, clusterId, configuration, LOG,
+            ozoneContainer.getDbVolumeSet());
+        if (!result) {
+          volumeSet.failVolume(volume.getStorageDir().getPath());
+        }
+      }
+      if (volumeSet.getVolumesList().size() == 0) {
+        // All volumes are in inconsistent state
+        throw new DiskOutOfSpaceException(
+            "All configured Volumes are in Inconsistent State");
+      }
+    } finally {
+      volumeSet.writeUnlock();
+    }
+  }
 }

hadoop-hdds/container-service/src/main/java/org/apache/hadoop/ozone/container/common/utils/DatanodeStoreCache.java

@@ -58,7 +58,17 @@ public RawDB getDB(String containerDBPath) {
   }
 
   public void removeDB(String containerDBPath) {
-    datanodeStoreMap.remove(containerDBPath);
+    RawDB db = datanodeStoreMap.remove(containerDBPath);
+    if (db == null) {
+      LOG.debug("DB {} already removed", containerDBPath);
+      return;
+    }
+
+    try {
+      db.getStore().stop();
+    } catch (Exception e) {
+      LOG.error("Stop DatanodeStore: {} failed", containerDBPath, e);
+    }
   }
 
   public void shutdownCache() {