HDDS-14041. Add metrics to track Snapshot RocksDB space and SST File stats.

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

@@ -2055,6 +2055,15 @@
       file. Unit could be defined with postfix (ns,ms,s,m,h,d)
     </description>
   </property>
+  <property>
+    <name>ozone.om.snapshot.directory.metrics.update.interval</name>
+    <value>5m</value>
+    <tag>OZONE, OM</tag>
+    <description>Time interval used to update the space consumption stats of the
+      Ozone Manager snapshot directories. Background thread periodically saves these
+      stats. Unit could be defined with postfix (ns,ms,s,m,h,d)
+    </description>
+  </property>
   <property>
     <name>ozone.security.enabled</name>
     <value>false</value>

hadoop-ozone/common/src/main/java/org/apache/hadoop/ozone/om/OMConfigKeys.java

@@ -167,6 +167,10 @@ public final class OMConfigKeys {
   public static final boolean
       OZONE_OM_SNAPSHOT_ROCKSDB_METRICS_ENABLED_DEFAULT = false;
 
+  public static final String OZONE_OM_SNAPSHOT_DIRECTORY_METRICS_UPDATE_INTERVAL =
+      "ozone.om.snapshot.directory.metrics.update.interval";
+  public static final String OZONE_OM_SNAPSHOT_DIRECTORY_METRICS_UPDATE_INTERVAL_DEFAULT = "5m";
+
   /**
    * OM Ratis related configurations.
    */

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

@@ -19,13 +19,15 @@
 
 import com.google.common.annotations.VisibleForTesting;
 import org.apache.hadoop.hdds.annotation.InterfaceAudience;
+import org.apache.hadoop.hdds.conf.OzoneConfiguration;
 import org.apache.hadoop.hdds.utils.DBCheckpointMetrics;
 import org.apache.hadoop.metrics2.MetricsSystem;
 import org.apache.hadoop.metrics2.annotation.Metric;
 import org.apache.hadoop.metrics2.annotation.Metrics;
 import org.apache.hadoop.metrics2.lib.DefaultMetricsSystem;
 import org.apache.hadoop.metrics2.lib.MutableCounterLong;
 import org.apache.hadoop.metrics2.lib.MutableGaugeInt;
+import org.apache.hadoop.ozone.om.snapshot.OMSnapshotDirectoryMetrics;
 
 /**
  * This class is for maintaining Ozone Manager statistics.
@@ -245,6 +247,7 @@ public class OMMetrics implements OmMetadataReaderMetrics {
   private @Metric MutableCounterLong ecBucketCreateTotal;
   private @Metric MutableCounterLong ecBucketCreateFailsTotal;
   private final DBCheckpointMetrics dbCheckpointMetrics;
+  private OMSnapshotDirectoryMetrics snapshotDirectoryMetrics;
 
   public OMMetrics() {
     dbCheckpointMetrics = DBCheckpointMetrics.create("OM Metrics");
@@ -261,6 +264,39 @@ public DBCheckpointMetrics getDBCheckpointMetrics() {
     return dbCheckpointMetrics;
   }
 
+  /**
+   * @return an instance of OMSnapshotDirectoryMetrics.
+   * @throws IllegalStateException if the metric is null
+   */
+  public OMSnapshotDirectoryMetrics getSnapshotDirectoryMetrics() throws IllegalStateException {
+    if (snapshotDirectoryMetrics == null) {
+      throw new IllegalStateException(
+          "SnapshotDirectoryMetrics not initialized. Call startSnapshotDirectoryMetrics() first.");
+    }
+    return snapshotDirectoryMetrics;
+  }
+
+  /**
+   * Starts periodic updates for snapshot directory metrics.
+   * Creates the metrics instance if it doesn't exist.
+   *
+   * @param configuration OzoneConfiguration for reading update interval
+   * @param metadataManager OMMetadataManager for accessing snapshot directories
+   */
+  public void startSnapshotDirectoryMetrics(OzoneConfiguration configuration,
+      OMMetadataManager metadataManager) {
+    if (snapshotDirectoryMetrics == null) {
+      snapshotDirectoryMetrics = OMSnapshotDirectoryMetrics.create("OM Metrics", metadataManager);
+    }
+    snapshotDirectoryMetrics.start(configuration);
+  }
+
+  public void stopSnapshotDirectoryMetrics() {
+    if (snapshotDirectoryMetrics != null) {
+      snapshotDirectoryMetrics.stop();
+    }
+  }
+
   public void incNumS3BucketCreates() {
     numBucketOps.incr();
     numBucketS3Creates.incr();

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

@@ -1868,6 +1868,10 @@ public void start() throws IOException {
     metricsTimer = new Timer();
     metricsTimer.schedule(scheduleOMMetricsWriteTask, 0, period);
 
+    LOG.info("Starting OmSnapshotDirectoryMetrics");
+
+    metrics.startSnapshotDirectoryMetrics(configuration, getMetadataManager());
+
     try {
       scmTopologyClient.start(configuration);
     } catch (IOException ex) {
@@ -1950,6 +1954,10 @@ public void restart() throws IOException {
     metricsTimer = new Timer();
     metricsTimer.schedule(scheduleOMMetricsWriteTask, 0, period);
 
+    // Restart snapshot directory metrics updates
+    metrics.stopSnapshotDirectoryMetrics();
+    metrics.startSnapshotDirectoryMetrics(configuration, getMetadataManager());
+
     initializeRatisServer(false);
     if (omRatisServer != null) {
       omRatisServer.start();
@@ -2425,6 +2433,7 @@ public boolean stop() {
       stopTrashEmptier();
       metadataManager.stop();
       omSnapshotManager.close();
+      metrics.stopSnapshotDirectoryMetrics();
       metrics.unRegister();
       omClientProtocolMetrics.unregister();
       unregisterMXBean();