[server] Extend metadata models to support per-table/partition remote data directory

Author: LiebingYu

fluss-client/src/main/java/org/apache/fluss/client/admin/FlussAdmin.java

@@ -298,6 +298,9 @@ public CompletableFuture<TableInfo> getTableInfo(TablePath tablePath) {
                                         r.getTableId(),
                                         r.getSchemaId(),
                                         TableDescriptor.fromJsonBytes(r.getTableJson()),
+                                        // For backward compatibility, results returned by old
+                                        // clusters do not include the remote data dir
+                                        r.hasRemoteDataDir() ? r.getRemoteDataDir() : null,
                                         r.getCreatedTime(),
                                         r.getModifiedTime()));
     }

fluss-client/src/main/java/org/apache/fluss/client/utils/ClientRpcMessageUtils.java

@@ -555,8 +555,12 @@ public static List<PartitionInfo> toPartitionInfos(ListPartitionInfosResponse re
                         pbPartitionInfo ->
                                 new PartitionInfo(
                                         pbPartitionInfo.getPartitionId(),
-                                        toResolvedPartitionSpec(
-                                                pbPartitionInfo.getPartitionSpec())))
+                                        toResolvedPartitionSpec(pbPartitionInfo.getPartitionSpec()),
+                                        // For backward compatibility, results returned by old
+                                        // clusters do not include the remote data dir
+                                        pbPartitionInfo.hasRemoteDataDir()
+                                                ? pbPartitionInfo.getRemoteDataDir()
+                                                : null))
                 .collect(Collectors.toList());
     }
 

fluss-client/src/test/java/org/apache/fluss/client/table/scanner/log/DefaultCompletedFetchTest.java

@@ -67,6 +67,7 @@
 import static org.apache.fluss.record.TestData.DATA2_TABLE_ID;
 import static org.apache.fluss.record.TestData.DATA2_TABLE_INFO;
 import static org.apache.fluss.record.TestData.DATA2_TABLE_PATH;
+import static org.apache.fluss.record.TestData.DEFAULT_REMOTE_DATA_DIR;
 import static org.apache.fluss.record.TestData.DEFAULT_SCHEMA_ID;
 import static org.apache.fluss.row.BinaryString.fromString;
 import static org.apache.fluss.rpc.util.CommonRpcMessageUtils.toByteBuffer;
@@ -169,6 +170,7 @@ void testProjection(LogFormat logFormat, byte magic) throws Exception {
                                 .distributedBy(3)
                                 .logFormat(logFormat)
                                 .build(),
+                        DEFAULT_REMOTE_DATA_DIR,
                         System.currentTimeMillis(),
                         System.currentTimeMillis());
         long fetchOffset = 0L;
@@ -313,6 +315,7 @@ void testComplexTypeFetch() throws Exception {
                                 .distributedBy(3)
                                 .logFormat(LogFormat.ARROW)
                                 .build(),
+                        DEFAULT_REMOTE_DATA_DIR,
                         System.currentTimeMillis(),
                         System.currentTimeMillis());
         long fetchOffset = 0L;

fluss-client/src/test/java/org/apache/fluss/client/table/scanner/log/LogFetcherITCase.java

@@ -172,6 +172,7 @@ void testFetchWithSchemaChange() throws Exception {
                                 DATA1_TABLE_INFO.getNumBuckets(),
                                 DATA1_TABLE_INFO.getProperties(),
                                 DATA1_TABLE_INFO.getCustomProperties(),
+                                DATA1_TABLE_INFO.getRemoteDataDir(),
                                 DATA1_TABLE_INFO.getComment().orElse(null),
                                 DATA1_TABLE_INFO.getCreatedTime(),
                                 DATA1_TABLE_INFO.getModifiedTime()),

fluss-client/src/test/java/org/apache/fluss/client/table/scanner/log/RemoteCompletedFetchTest.java

@@ -61,6 +61,7 @@
 import static org.apache.fluss.record.TestData.DATA2_TABLE_ID;
 import static org.apache.fluss.record.TestData.DATA2_TABLE_INFO;
 import static org.apache.fluss.record.TestData.DATA2_TABLE_PATH;
+import static org.apache.fluss.record.TestData.DEFAULT_REMOTE_DATA_DIR;
 import static org.apache.fluss.record.TestData.DEFAULT_SCHEMA_ID;
 import static org.apache.fluss.testutils.DataTestUtils.genLogFile;
 import static org.apache.fluss.utils.FlussPaths.remoteLogSegmentDir;
@@ -218,6 +219,7 @@ void testProjection(String format) throws Exception {
                                 .distributedBy(3)
                                 .logFormat(logFormat)
                                 .build(),
+                        DEFAULT_REMOTE_DATA_DIR,
                         System.currentTimeMillis(),
                         System.currentTimeMillis());
         long fetchOffset = 0L;

fluss-client/src/test/java/org/apache/fluss/client/write/RecordAccumulatorTest.java

@@ -76,6 +76,7 @@
 import static org.apache.fluss.record.TestData.DATA1_TABLE_ID;
 import static org.apache.fluss.record.TestData.DATA1_TABLE_INFO;
 import static org.apache.fluss.record.TestData.DATA1_TABLE_PATH;
+import static org.apache.fluss.record.TestData.DEFAULT_REMOTE_DATA_DIR;
 import static org.apache.fluss.testutils.DataTestUtils.indexedRow;
 import static org.apache.fluss.testutils.DataTestUtils.row;
 import static org.assertj.core.api.Assertions.assertThat;
@@ -96,6 +97,7 @@ class RecordAccumulatorTest {
                             .distributedBy(3)
                             .property(ConfigOptions.TABLE_LOG_ARROW_COMPRESSION_TYPE.key(), "zstd")
                             .build(),
+                    DEFAULT_REMOTE_DATA_DIR,
                     System.currentTimeMillis(),
                     System.currentTimeMillis());
 

fluss-common/src/main/java/org/apache/fluss/config/FlussConfigUtils.java

@@ -59,6 +59,25 @@ public static boolean isAlterableTableOption(String key) {
         return ALTERABLE_TABLE_OPTIONS.contains(key);
     }
 
+    /**
+     * Returns the default remote data directory from the configuration. Used as a fallback for
+     * tables or partitions that do not contain remote data directory metadata.
+     *
+     * @param conf the Fluss configuration
+     * @return the default remote data directory path, never {@code null} if the configuration is
+     *     valid (i.e., at least one of {@code remote.data.dir} or {@code remote.data.dirs} is set)
+     * @see ConfigOptions#REMOTE_DATA_DIR
+     * @see ConfigOptions#REMOTE_DATA_DIRS
+     */
+    public static String getDefaultRemoteDataDir(Configuration conf) {
+        List<String> remoteDataDirs = conf.get(ConfigOptions.REMOTE_DATA_DIRS);
+        if (!remoteDataDirs.isEmpty()) {
+            return remoteDataDirs.get(0);
+        }
+
+        return conf.get(ConfigOptions.REMOTE_DATA_DIR);
+    }
+
     @VisibleForTesting
     static Map<String, ConfigOption<?>> extractConfigOptions(String prefix) {
         Map<String, ConfigOption<?>> options = new HashMap<>();

fluss-common/src/main/java/org/apache/fluss/metadata/PartitionInfo.java

@@ -22,19 +22,22 @@
 import java.util.Objects;
 
 /**
- * Information of a partition metadata, includes the partition's name and the partition id that
- * represents the unique identifier of the partition.
+ * Information of a partition metadata, includes partition id (unique identifier of the partition),
+ * partition name, remote data dir for partitioned data storage, etc.
  *
  * @since 0.2
  */
 @PublicEvolving
 public class PartitionInfo {
     private final long partitionId;
     private final ResolvedPartitionSpec partitionSpec;
+    private final String remoteDataDir;
 
-    public PartitionInfo(long partitionId, ResolvedPartitionSpec partitionSpec) {
+    public PartitionInfo(
+            long partitionId, ResolvedPartitionSpec partitionSpec, String remoteDataDir) {
         this.partitionId = partitionId;
         this.partitionSpec = partitionSpec;
+        this.remoteDataDir = remoteDataDir;
     }
 
     /** Get the partition id. The id is globally unique in the Fluss cluster. */
@@ -58,6 +61,10 @@ public PartitionSpec getPartitionSpec() {
         return partitionSpec.toPartitionSpec();
     }
 
+    public String getRemoteDataDir() {
+        return remoteDataDir;
+    }
+
     @Override
     public boolean equals(Object o) {
         if (this == o) {
@@ -67,16 +74,25 @@ public boolean equals(Object o) {
             return false;
         }
         PartitionInfo that = (PartitionInfo) o;
-        return partitionId == that.partitionId && Objects.equals(partitionSpec, that.partitionSpec);
+        return partitionId == that.partitionId
+                && Objects.equals(partitionSpec, that.partitionSpec)
+                && Objects.equals(remoteDataDir, that.remoteDataDir);
     }
 
     @Override
     public int hashCode() {
-        return Objects.hash(partitionId, partitionSpec);
+        return Objects.hash(partitionId, partitionSpec, remoteDataDir);
     }
 
     @Override
     public String toString() {
-        return "Partition{name='" + getPartitionName() + '\'' + ", id=" + partitionId + '}';
+        return "Partition{name='"
+                + getPartitionName()
+                + '\''
+                + ", id="
+                + partitionId
+                + ", remoteDataDir="
+                + remoteDataDir
+                + '}';
     }
 }

fluss-common/src/main/java/org/apache/fluss/metadata/TableInfo.java

@@ -59,6 +59,7 @@ public final class TableInfo {
     private final Configuration properties;
     private final TableConfig tableConfig;
     private final Configuration customProperties;
+    private final String remoteDataDir;
     private final @Nullable String comment;
 
     private final long createdTime;
@@ -74,6 +75,7 @@ public TableInfo(
             int numBuckets,
             Configuration properties,
             Configuration customProperties,
+            String remoteDataDir,
             @Nullable String comment,
             long createdTime,
             long modifiedTime) {
@@ -90,6 +92,7 @@ public TableInfo(
         this.properties = properties;
         this.tableConfig = new TableConfig(properties);
         this.customProperties = customProperties;
+        this.remoteDataDir = remoteDataDir;
         this.comment = comment;
         this.createdTime = createdTime;
         this.modifiedTime = modifiedTime;
@@ -263,6 +266,11 @@ public Configuration getCustomProperties() {
         return customProperties;
     }
 
+    /** Returns the remote data directory of the table. */
+    public String getRemoteDataDir() {
+        return remoteDataDir;
+    }
+
     /** Returns the comment/description of the table. */
     public Optional<String> getComment() {
         return Optional.ofNullable(comment);
@@ -314,6 +322,7 @@ public static TableInfo of(
             long tableId,
             int schemaId,
             TableDescriptor tableDescriptor,
+            String remoteDataDir,
             long createdTime,
             long modifiedTime) {
         Schema schema = tableDescriptor.getSchema();
@@ -335,6 +344,7 @@ public static TableInfo of(
                 numBuckets,
                 Configuration.fromMap(tableDescriptor.getProperties()),
                 Configuration.fromMap(tableDescriptor.getCustomProperties()),
+                remoteDataDir,
                 tableDescriptor.getComment().orElse(null),
                 createdTime,
                 modifiedTime);
@@ -358,6 +368,7 @@ public boolean equals(Object o) {
                 && Objects.equals(partitionKeys, that.partitionKeys)
                 && Objects.equals(properties, that.properties)
                 && Objects.equals(customProperties, that.customProperties)
+                && Objects.equals(remoteDataDir, that.remoteDataDir)
                 && Objects.equals(comment, that.comment);
     }
 
@@ -376,6 +387,7 @@ public int hashCode() {
                 numBuckets,
                 properties,
                 customProperties,
+                remoteDataDir,
                 comment);
     }
 
@@ -402,6 +414,8 @@ public String toString() {
                 + properties
                 + ", customProperties="
                 + customProperties
+                + ", remoteDataDir="
+                + remoteDataDir
                 + ", comment='"
                 + comment
                 + '\''

fluss-common/src/test/java/org/apache/fluss/record/TestData.java

@@ -39,6 +39,7 @@ public final class TestData {
     public static final short DEFAULT_SCHEMA_ID = 1;
     public static final long BASE_OFFSET = 0L;
     public static final byte DEFAULT_MAGIC = CURRENT_LOG_MAGIC_VALUE;
+    public static final String DEFAULT_REMOTE_DATA_DIR = "/tmp/fluss/remote-data";
     // ---------------------------- data1 and related table info begin ---------------------------
     public static final List<Object[]> DATA1 =
             Arrays.asList(
@@ -93,6 +94,7 @@ public final class TestData {
                     DATA1_TABLE_ID,
                     1,
                     DATA1_TABLE_DESCRIPTOR,
+                    DEFAULT_REMOTE_DATA_DIR,
                     currentMillis,
                     currentMillis);
 
@@ -118,6 +120,7 @@ public final class TestData {
                     PARTITION_TABLE_ID,
                     1,
                     DATA1_PARTITIONED_TABLE_DESCRIPTOR,
+                    DEFAULT_REMOTE_DATA_DIR,
                     System.currentTimeMillis(),
                     System.currentTimeMillis());
 
@@ -148,6 +151,7 @@ public final class TestData {
                     DATA1_TABLE_ID_PK,
                     1,
                     DATA1_TABLE_DESCRIPTOR_PK,
+                    DEFAULT_REMOTE_DATA_DIR,
                     currentMillis,
                     currentMillis);
 
@@ -217,6 +221,7 @@ public final class TestData {
                     DATA2_TABLE_ID,
                     1,
                     DATA2_TABLE_DESCRIPTOR,
+                    DEFAULT_REMOTE_DATA_DIR,
                     System.currentTimeMillis(),
                     System.currentTimeMillis());
     // -------------------------------- data2 info end ------------------------------------