[HUDI-1381] Schedule compaction based on time elapsed

hudi-client/hudi-client-common/src/main/java/org/apache/hudi/config/HoodieCompactionConfig.java

@@ -46,6 +46,9 @@ public class HoodieCompactionConfig extends DefaultHoodieConfig {
   public static final String INLINE_COMPACT_PROP = "hoodie.compact.inline";
   // Run a compaction every N delta commits
   public static final String INLINE_COMPACT_NUM_DELTA_COMMITS_PROP = "hoodie.compact.inline.max.delta.commits";
+  public static final String INLINE_COMPACT_ELAPSED_TIME_PROP = "hoodie.compact.inline.max.delta.time";
+  public static final String INLINE_COMPACT_NUM_DELTA_COMMITS_ENABLED_PROP = "hoodie.compact.inline.max.delta.commits.enable";
+  public static final String INLINE_COMPACT_ELAPSED_TIME_ENABLED_PROP = "hoodie.compact.inline.max.delta.time.enable";
   public static final String CLEANER_FILE_VERSIONS_RETAINED_PROP = "hoodie.cleaner.fileversions.retained";
   public static final String CLEANER_COMMITS_RETAINED_PROP = "hoodie.cleaner.commits.retained";
   public static final String CLEANER_INCREMENTAL_MODE = "hoodie.cleaner.incremental.mode";
@@ -109,6 +112,9 @@ public class HoodieCompactionConfig extends DefaultHoodieConfig {
   private static final String DEFAULT_INLINE_COMPACT = "false";
   private static final String DEFAULT_INCREMENTAL_CLEANER = "true";
   private static final String DEFAULT_INLINE_COMPACT_NUM_DELTA_COMMITS = "5";
+  private static final String DEFAULT_INLINE_COMPACT_ELAPSED_TIME = String.valueOf(60 * 60);
+  private static final String DEFAULT_INLINE_COMPACT_NUM_DELTA_COMMITS_ENABLED = "true";
+  private static final String DEFAULT_INLINE_COMPACT_ELAPSED_TIME_ENABLED = "false";
   private static final String DEFAULT_CLEANER_FILE_VERSIONS_RETAINED = "3";
   private static final String DEFAULT_CLEANER_COMMITS_RETAINED = "10";
   private static final String DEFAULT_MAX_COMMITS_TO_KEEP = "30";
@@ -235,6 +241,21 @@ public Builder withMaxNumDeltaCommitsBeforeCompaction(int maxNumDeltaCommitsBefo
       return this;
     }
 
+    public Builder withMaxDeltaTimeBeforeCompaction(int maxDeltaTimeBeforeCompaction) {
+      props.setProperty(INLINE_COMPACT_ELAPSED_TIME_PROP, String.valueOf(maxDeltaTimeBeforeCompaction));
+      return this;
+    }
+
+    public Builder withMaxNumDeltaCommitsBeforeCompactionEnabled(boolean maxNumDeltaCommitsBeforeCompactionEnabled) {
+      props.setProperty(INLINE_COMPACT_NUM_DELTA_COMMITS_ENABLED_PROP, String.valueOf(maxNumDeltaCommitsBeforeCompactionEnabled));
+      return this;
+    }
+
+    public Builder withMaxDeltaTimeBeforeCompactionEnabled(boolean withMaxDeltaTimeBeforeCompactionEnabled) {
+      props.setProperty(INLINE_COMPACT_ELAPSED_TIME_ENABLED_PROP, String.valueOf(withMaxDeltaTimeBeforeCompactionEnabled));
+      return this;
+    }
+
     public Builder withCompactionLazyBlockReadEnabled(Boolean compactionLazyBlockReadEnabled) {
       props.setProperty(COMPACTION_LAZY_BLOCK_READ_ENABLED_PROP, String.valueOf(compactionLazyBlockReadEnabled));
       return this;
@@ -271,6 +292,12 @@ public HoodieCompactionConfig build() {
           DEFAULT_INLINE_COMPACT);
       setDefaultOnCondition(props, !props.containsKey(INLINE_COMPACT_NUM_DELTA_COMMITS_PROP),
           INLINE_COMPACT_NUM_DELTA_COMMITS_PROP, DEFAULT_INLINE_COMPACT_NUM_DELTA_COMMITS);
+      setDefaultOnCondition(props, !props.containsKey(INLINE_COMPACT_ELAPSED_TIME_PROP),
+          INLINE_COMPACT_ELAPSED_TIME_PROP, DEFAULT_INLINE_COMPACT_ELAPSED_TIME);
+      setDefaultOnCondition(props, !props.containsKey(INLINE_COMPACT_ELAPSED_TIME_ENABLED_PROP),
+          INLINE_COMPACT_ELAPSED_TIME_ENABLED_PROP, DEFAULT_INLINE_COMPACT_ELAPSED_TIME_ENABLED);
+      setDefaultOnCondition(props, !props.containsKey(INLINE_COMPACT_NUM_DELTA_COMMITS_ENABLED_PROP),
+          INLINE_COMPACT_NUM_DELTA_COMMITS_ENABLED_PROP, DEFAULT_INLINE_COMPACT_NUM_DELTA_COMMITS_ENABLED);
       setDefaultOnCondition(props, !props.containsKey(CLEANER_POLICY_PROP), CLEANER_POLICY_PROP,
           DEFAULT_CLEANER_POLICY);
       setDefaultOnCondition(props, !props.containsKey(CLEANER_FILE_VERSIONS_RETAINED_PROP),

hudi-client/hudi-client-common/src/main/java/org/apache/hudi/config/HoodieWriteConfig.java

@@ -18,6 +18,7 @@
 
 package org.apache.hudi.config;
 
+import org.apache.hadoop.hbase.io.compress.Compression;
 import org.apache.hudi.client.WriteStatus;
 import org.apache.hudi.client.bootstrap.BootstrapMode;
 import org.apache.hudi.common.config.DefaultHoodieConfig;
@@ -36,7 +37,6 @@
 import org.apache.hudi.metrics.datadog.DatadogHttpClient.ApiSite;
 import org.apache.hudi.table.action.compact.strategy.CompactionStrategy;
 
-import org.apache.hadoop.hbase.io.compress.Compression;
 import org.apache.parquet.hadoop.metadata.CompressionCodecName;
 
 import javax.annotation.concurrent.Immutable;
@@ -397,10 +397,22 @@ public boolean isInlineCompaction() {
     return Boolean.parseBoolean(props.getProperty(HoodieCompactionConfig.INLINE_COMPACT_PROP));
   }
 
+  public boolean getInlineCompactDeltaElapsedEnabled() {
+    return Boolean.parseBoolean(props.getProperty(HoodieCompactionConfig.INLINE_COMPACT_ELAPSED_TIME_ENABLED_PROP));
+  }
+
+  public boolean getInlineCompactDeltaNumCommitEnabled() {
+    return Boolean.parseBoolean(props.getProperty(HoodieCompactionConfig.INLINE_COMPACT_NUM_DELTA_COMMITS_ENABLED_PROP));
+  }
+
   public int getInlineCompactDeltaCommitMax() {
     return Integer.parseInt(props.getProperty(HoodieCompactionConfig.INLINE_COMPACT_NUM_DELTA_COMMITS_PROP));
   }
 
+  public int getInlineCompactDeltaElapsedTimeMax() {
+    return Integer.parseInt(props.getProperty(HoodieCompactionConfig.INLINE_COMPACT_ELAPSED_TIME_PROP));
+  }
+
   public CompactionStrategy getCompactionStrategy() {
     return ReflectionUtils.loadClass(props.getProperty(HoodieCompactionConfig.COMPACTION_STRATEGY_PROP));
   }
@@ -468,7 +480,7 @@ public String getClusteringExecutionStrategyClass() {
   public long getClusteringMaxBytesInGroup() {
     return Long.parseLong(props.getProperty(HoodieClusteringConfig.CLUSTERING_MAX_BYTES_PER_GROUP));
   }
-  
+
   public long getClusteringSmallFileLimit() {
     return Long.parseLong(props.getProperty(HoodieClusteringConfig.CLUSTERING_PLAN_SMALL_FILE_LIMIT));
   }
@@ -484,7 +496,7 @@ public long getClusteringTargetFileMaxBytes() {
   public int getTargetPartitionsForClustering() {
     return Integer.parseInt(props.getProperty(HoodieClusteringConfig.CLUSTERING_TARGET_PARTITIONS));
   }
-  
+
   public String getClusteringSortColumns() {
     return props.getProperty(HoodieClusteringConfig.CLUSTERING_SORT_COLUMNS_PROPERTY);
   }

hudi-client/hudi-spark-client/src/main/java/org/apache/hudi/table/action/compact/SparkScheduleCompactionActionExecutor.java

@@ -25,7 +25,9 @@
 import org.apache.hudi.common.model.HoodieKey;
 import org.apache.hudi.common.model.HoodieRecord;
 import org.apache.hudi.common.model.HoodieRecordPayload;
+import org.apache.hudi.common.table.timeline.HoodieActiveTimeline;
 import org.apache.hudi.common.table.timeline.HoodieInstant;
+import org.apache.hudi.common.table.timeline.HoodieTimeline;
 import org.apache.hudi.common.table.view.SyncableFileSystemView;
 import org.apache.hudi.common.util.Option;
 import org.apache.hudi.common.util.collection.Pair;
@@ -37,6 +39,7 @@
 import org.apache.spark.api.java.JavaRDD;
 
 import java.io.IOException;
+import java.text.ParseException;
 import java.util.Map;
 import java.util.Set;
 import java.util.stream.Collectors;
@@ -60,17 +63,32 @@ protected HoodieCompactionPlan scheduleCompaction() {
     LOG.info("Checking if compaction needs to be run on " + config.getBasePath());
     Option<HoodieInstant> lastCompaction = table.getActiveTimeline().getCommitTimeline()
         .filterCompletedInstants().lastInstant();
-    String lastCompactionTs = "0";
+    HoodieTimeline deltaCommits = table.getActiveTimeline().getDeltaCommitTimeline();
+    String lastCompactionTs;
+    int deltaCommitsSinceLastCompaction;
     if (lastCompaction.isPresent()) {
       lastCompactionTs = lastCompaction.get().getTimestamp();
+      deltaCommitsSinceLastCompaction = deltaCommits.findInstantsAfter(lastCompactionTs, Integer.MAX_VALUE).countInstants();
+    } else {
+      lastCompactionTs = deltaCommits.firstInstant().get().getTimestamp();
+      deltaCommitsSinceLastCompaction = deltaCommits.findInstantsAfterOrEquals(lastCompactionTs, Integer.MAX_VALUE).countInstants();
     }
-
-    int deltaCommitsSinceLastCompaction = table.getActiveTimeline().getDeltaCommitTimeline()
-        .findInstantsAfter(lastCompactionTs, Integer.MAX_VALUE).countInstants();
-    if (config.getInlineCompactDeltaCommitMax() > deltaCommitsSinceLastCompaction) {
-      LOG.info("Not scheduling compaction as only " + deltaCommitsSinceLastCompaction
-          + " delta commits was found since last compaction " + lastCompactionTs + ". Waiting for "
-          + config.getInlineCompactDeltaCommitMax());
+    // judge if we need to compact according to num delta commits and time elapsed
+    boolean numCommitEnabled = config.getInlineCompactDeltaNumCommitEnabled();
+    boolean timeEnabled = config.getInlineCompactDeltaElapsedEnabled();
+    boolean compactable;
+    if (numCommitEnabled && !timeEnabled) {
+      compactable = config.getInlineCompactDeltaCommitMax() > deltaCommitsSinceLastCompaction;
+    } else if (!numCommitEnabled && timeEnabled) {
+      compactable = parseToTimestamp(lastCompactionTs) + config.getInlineCompactDeltaElapsedTimeMax() > parseToTimestamp(instantTime);
+    } else {
+      compactable = config.getInlineCompactDeltaCommitMax() > deltaCommitsSinceLastCompaction
+          && parseToTimestamp(lastCompactionTs) + config.getInlineCompactDeltaElapsedTimeMax() > parseToTimestamp(instantTime);
+    }
+    if (compactable) {
+      LOG.info(String.format("Not scheduling compaction as only %s delta commits was found since last compaction %s."
+              + "Waiting for %s,or %sms elapsed time need since last compaction %s.", deltaCommitsSinceLastCompaction,
+          lastCompactionTs, config.getInlineCompactDeltaCommitMax(), config.getInlineCompactDeltaElapsedTimeMax(), lastCompactionTs));
       return new HoodieCompactionPlan();
     }
 
@@ -90,4 +108,13 @@ protected HoodieCompactionPlan scheduleCompaction() {
     }
   }
 
+  public Long parseToTimestamp(String time) {
+    long timestamp;
+    try {
+      timestamp = HoodieActiveTimeline.COMMIT_FORMATTER.parse(time).getTime() / 1000;
+    } catch (ParseException e) {
+      throw new HoodieCompactionException(e.getMessage(), e);
+    }
+    return timestamp;
+  }
 }

hudi-client/hudi-spark-client/src/test/java/org/apache/hudi/table/action/compact/TestInlineCompaction.java

@@ -24,7 +24,6 @@
 import org.apache.hudi.common.table.HoodieTableMetaClient;
 import org.apache.hudi.common.table.timeline.HoodieActiveTimeline;
 import org.apache.hudi.common.table.timeline.HoodieTimeline;
-import org.apache.hudi.common.util.CollectionUtils;
 import org.apache.hudi.config.HoodieCompactionConfig;
 import org.apache.hudi.config.HoodieWriteConfig;
 import org.junit.jupiter.api.Test;
@@ -39,21 +38,25 @@
 
 public class TestInlineCompaction extends CompactionTestBase {
 
-  private HoodieWriteConfig getConfigForInlineCompaction(int maxDeltaCommits) {
+  private HoodieWriteConfig getConfigForInlineCompaction(int maxDeltaCommits, int maxDeltaTime) {
     return getConfigBuilder(false)
         .withCompactionConfig(HoodieCompactionConfig.newBuilder()
-            .withInlineCompaction(true).withMaxNumDeltaCommitsBeforeCompaction(maxDeltaCommits).build())
+            .withInlineCompaction(true)
+            .withMaxNumDeltaCommitsBeforeCompaction(maxDeltaCommits)
+            .withMaxDeltaTimeBeforeCompaction(maxDeltaTime)
+            .withMaxDeltaTimeBeforeCompactionEnabled(true).build())
         .build();
   }
 
   @Test
   public void testCompactionIsNotScheduledEarly() throws Exception {
     // Given: make two commits
-    HoodieWriteConfig cfg = getConfigForInlineCompaction(3);
+    HoodieWriteConfig cfg = getConfigForInlineCompaction(3, 60);
     try (SparkRDDWriteClient<?> writeClient = getHoodieWriteClient(cfg)) {
-      List<HoodieRecord> records = dataGen.generateInserts("000", 100);
+      List<HoodieRecord> records = dataGen.generateInserts(HoodieActiveTimeline.createNewInstantTime(), 100);
       HoodieReadClient readClient = getHoodieReadClient(cfg.getBasePath());
-      runNextDeltaCommits(writeClient, readClient, Arrays.asList("000", "001"), records, cfg, true, new ArrayList<>());
+      List<String> instants = IntStream.range(0, 2).mapToObj(i -> HoodieActiveTimeline.createNewInstantTime()).collect(Collectors.toList());
+      runNextDeltaCommits(writeClient, readClient, instants, records, cfg, true, new ArrayList<>());
       HoodieTableMetaClient metaClient = new HoodieTableMetaClient(hadoopConf, cfg.getBasePath());
 
       // Then: ensure no compaction is executedm since there are only 2 delta commits
@@ -62,9 +65,9 @@ public void testCompactionIsNotScheduledEarly() throws Exception {
   }
 
   @Test
-  public void testSuccessfulCompaction() throws Exception {
+  public void testSuccessfulCompactionBasedOnNumCommits() throws Exception {
     // Given: make three commits
-    HoodieWriteConfig cfg = getConfigForInlineCompaction(3);
+    HoodieWriteConfig cfg = getConfigForInlineCompaction(3, 60);
     List<String> instants = IntStream.range(0, 2).mapToObj(i -> HoodieActiveTimeline.createNewInstantTime()).collect(Collectors.toList());
 
     try (SparkRDDWriteClient<?> writeClient = getHoodieWriteClient(cfg)) {
@@ -85,32 +88,95 @@ public void testSuccessfulCompaction() throws Exception {
   }
 
   @Test
-  public void testCompactionRetryOnFailure() throws Exception {
+  public void testSuccessfulCompactionBasedOnTime() throws Exception {
+    // Given: make one commit
+    HoodieWriteConfig cfg = getConfigForInlineCompaction(5, 10);
+
+    try (SparkRDDWriteClient<?> writeClient = getHoodieWriteClient(cfg)) {
+      String instantTime = HoodieActiveTimeline.createNewInstantTime();
+      List<HoodieRecord> records = dataGen.generateInserts(instantTime, 10);
+      HoodieReadClient readClient = getHoodieReadClient(cfg.getBasePath());
+      runNextDeltaCommits(writeClient, readClient, Arrays.asList(instantTime), records, cfg, true, new ArrayList<>());
+
+      // after 10s, that will trigger compaction
+      Thread.sleep(10000);
+      String finalInstant = HoodieActiveTimeline.createNewInstantTime();
+      HoodieTableMetaClient metaClient = new HoodieTableMetaClient(hadoopConf, cfg.getBasePath());
+      createNextDeltaCommit(finalInstant, dataGen.generateUpdates(finalInstant, 100), writeClient, metaClient, cfg, false);
+
+      // Then: ensure the file slices are compacted as per policy
+      metaClient = new HoodieTableMetaClient(hadoopConf, cfg.getBasePath());
+      assertEquals(3, metaClient.getActiveTimeline().getCommitsAndCompactionTimeline().countInstants());
+      assertEquals(HoodieTimeline.COMMIT_ACTION, metaClient.getActiveTimeline().lastInstant().get().getAction());
+    }
+  }
+
+  @Test
+  public void testCompactionRetryOnFailureBasedOnNumCommits() throws Exception {
     // Given: two commits, schedule compaction and its failed/in-flight
     HoodieWriteConfig cfg = getConfigBuilder(false)
         .withCompactionConfig(HoodieCompactionConfig.newBuilder()
             .withInlineCompaction(false).withMaxNumDeltaCommitsBeforeCompaction(1).build())
         .build();
-    List<String> instants = CollectionUtils.createImmutableList("000", "001");
+    List<String> instants = IntStream.range(0, 2).mapToObj(i -> HoodieActiveTimeline.createNewInstantTime()).collect(Collectors.toList());
+    String instantTime2;
     try (SparkRDDWriteClient<?> writeClient = getHoodieWriteClient(cfg)) {
       List<HoodieRecord> records = dataGen.generateInserts(instants.get(0), 100);
       HoodieReadClient readClient = getHoodieReadClient(cfg.getBasePath());
       runNextDeltaCommits(writeClient, readClient, instants, records, cfg, true, new ArrayList<>());
-      // Schedule compaction 002, make it in-flight (simulates inline compaction failing)
-      scheduleCompaction("002", writeClient, cfg);
-      moveCompactionFromRequestedToInflight("002", cfg);
+      // Schedule compaction instant2, make it in-flight (simulates inline compaction failing)
+      instantTime2 = HoodieActiveTimeline.createNewInstantTime();
+      scheduleCompaction(instantTime2, writeClient, cfg);
+      moveCompactionFromRequestedToInflight(instantTime2, cfg);
     }
 
     // When: a third commit happens
-    HoodieWriteConfig inlineCfg = getConfigForInlineCompaction(2);
+    HoodieWriteConfig inlineCfg = getConfigForInlineCompaction(2, 60);
+    String instantTime3 = HoodieActiveTimeline.createNewInstantTime();
+    try (SparkRDDWriteClient<?> writeClient = getHoodieWriteClient(inlineCfg)) {
+      HoodieTableMetaClient metaClient = new HoodieTableMetaClient(hadoopConf, cfg.getBasePath());
+      createNextDeltaCommit(instantTime3, dataGen.generateUpdates(instantTime3, 100), writeClient, metaClient, inlineCfg, false);
+    }
+
+    // Then: 1 delta commit is done, the failed compaction is retried
+    metaClient = new HoodieTableMetaClient(hadoopConf, cfg.getBasePath());
+    assertEquals(4, metaClient.getActiveTimeline().getCommitsAndCompactionTimeline().countInstants());
+    assertEquals(instantTime2, metaClient.getActiveTimeline().getCommitTimeline().filterCompletedInstants().firstInstant().get().getTimestamp());
+  }
+
+  @Test
+  public void testCompactionRetryOnFailureBasedOnTime() throws Exception {
+    // Given: two commits, schedule compaction and its failed/in-flight
+    HoodieWriteConfig cfg = getConfigBuilder(false)
+        .withCompactionConfig(HoodieCompactionConfig.newBuilder()
+            .withInlineCompaction(false).withMaxDeltaTimeBeforeCompaction(5)
+            .withMaxDeltaTimeBeforeCompactionEnabled(true).build())
+        .build();
+    String instantTime;
+    List<String> instants = IntStream.range(0, 2).mapToObj(i -> HoodieActiveTimeline.createNewInstantTime()).collect(Collectors.toList());
+    try (SparkRDDWriteClient<?> writeClient = getHoodieWriteClient(cfg)) {
+      List<HoodieRecord> records = dataGen.generateInserts(instants.get(0), 100);
+      HoodieReadClient readClient = getHoodieReadClient(cfg.getBasePath());
+      runNextDeltaCommits(writeClient, readClient, instants, records, cfg, true, new ArrayList<>());
+      // Schedule compaction instantTime, make it in-flight (simulates inline compaction failing)
+      Thread.sleep(10000);
+      instantTime = HoodieActiveTimeline.createNewInstantTime();
+      scheduleCompaction(instantTime, writeClient, cfg);
+      moveCompactionFromRequestedToInflight(instantTime, cfg);
+    }
+
+    // When: commit happens after 10s
+    HoodieWriteConfig inlineCfg = getConfigForInlineCompaction(5, 10);
+    String instantTime2;
     try (SparkRDDWriteClient<?> writeClient = getHoodieWriteClient(inlineCfg)) {
       HoodieTableMetaClient metaClient = new HoodieTableMetaClient(hadoopConf, cfg.getBasePath());
-      createNextDeltaCommit("003", dataGen.generateUpdates("003", 100), writeClient, metaClient, inlineCfg, false);
+      instantTime2 = HoodieActiveTimeline.createNewInstantTime();
+      createNextDeltaCommit(instantTime2, dataGen.generateUpdates(instantTime2, 10), writeClient, metaClient, inlineCfg, false);
     }
 
     // Then: 1 delta commit is done, the failed compaction is retried
     metaClient = new HoodieTableMetaClient(hadoopConf, cfg.getBasePath());
     assertEquals(4, metaClient.getActiveTimeline().getCommitsAndCompactionTimeline().countInstants());
-    assertEquals("002", metaClient.getActiveTimeline().getCommitTimeline().filterCompletedInstants().firstInstant().get().getTimestamp());
+    assertEquals(instantTime, metaClient.getActiveTimeline().getCommitTimeline().filterCompletedInstants().firstInstant().get().getTimestamp());
   }
 }