Add more index options for streaming job (#76)

* Add watermark delay option Signed-off-by: Chen Dai <[email protected]> * Change UT and IT with watermark delay option Signed-off-by: Chen Dai <[email protected]> * Update MV SQL IT Signed-off-by: Chen Dai <[email protected]> * Add output mode option Signed-off-by: Chen Dai <[email protected]> * Add extra sink options Signed-off-by: Chen Dai <[email protected]> * Add extra source options Signed-off-by: Chen Dai <[email protected]> * Refactor MV UT Signed-off-by: Chen Dai <[email protected]> * Copy extra options to streaming relation Signed-off-by: Chen Dai <[email protected]> * Fix logical plan assert Signed-off-by: Chen Dai <[email protected]> * Qualify table name in relation when fetching extra option Signed-off-by: Chen Dai <[email protected]> * Update user manual with new options Signed-off-by: Chen Dai <[email protected]> --------- Signed-off-by: Chen Dai <[email protected]>
opensearch-project · Oct 21, 2023 · af210ca · af210ca
1 parent f8ec62e
commit af210ca
Show file tree

Hide file tree

Showing 10 changed files with 241 additions and 93 deletions.
diff --git a/docs/index.md b/docs/index.md
@@ -227,15 +227,22 @@ User can provide the following options in `WITH` clause of create statement:
 + `auto_refresh`: triggers Incremental Refresh immediately after index create complete if true. Otherwise, user has to trigger Full Refresh by `REFRESH` statement manually.
 + `refresh_interval`: a string as the time interval for incremental refresh, e.g. 1 minute, 10 seconds. This is only applicable when auto refresh enabled. Please check `org.apache.spark.unsafe.types.CalendarInterval` for valid duration identifiers. By default, next micro batch will be generated as soon as the previous one complete processing.
 + `checkpoint_location`: a string as the location path for incremental refresh job checkpoint. The location has to be a path in an HDFS compatible file system and only applicable when auto refresh enabled. If unspecified, temporary checkpoint directory will be used and may result in checkpoint data lost upon restart.
++ `watermark_delay`: a string as time expression for how late data can come and still be processed, e.g. 1 minute, 10 seconds. This is required by incremental refresh on materialized view if it has aggregation in the query.
++ `output_mode`: a mode string that describes how data will be written to streaming sink. If unspecified, default append mode will be applied.
 + `index_settings`: a JSON string as index settings for OpenSearch index that will be created. Please follow the format in OpenSearch documentation. If unspecified, default OpenSearch index settings will be applied.
++ `extra_options`: a JSON string as extra options that can be passed to Spark streaming source and sink API directly. Use qualified source table name (because there could be multiple) and "sink", e.g. '{"sink": "{key: val}", "table1": {key: val}}'
 
-Note that the index option name is case-sensitive.
+Note that the index option name is case-sensitive. Here is an example:
 
 ```sql
 WITH (
-  auto_refresh = [true|false],
-  refresh_interval = 'time interval expression',
-  checkpoint_location = 'checkpoint directory path'
+  auto_refresh = true,
+  refresh_interval = '10 Seconds',
+  checkpoint_location = 's3://test/',
+  watermark_delay = '1 Second',
+  output_mode = 'complete',
+  index_settings = '{"number_of_shards": 2, "number_of_replicas": 3}',
+  extra_options = '{"spark_catalog.default.alb_logs": {"maxFilesPerTrigger": "1"}}'
 )
 ```
 

diff --git a/flint-spark-integration/src/main/scala/org/opensearch/flint/spark/FlintSpark.scala b/flint-spark-integration/src/main/scala/org/opensearch/flint/spark/FlintSpark.scala
@@ -138,17 +138,18 @@ class FlintSpark(val spark: SparkSession) {
             .queryName(indexName)
             .format(FLINT_DATASOURCE)
             .options(flintSparkConf.properties)
-            .addIndexOptions(options)
+            .addSinkOptions(options)
             .start(indexName)
         Some(job.id.toString)
 
       // Otherwise, fall back to foreachBatch + batch refresh
       case INCREMENTAL =>
         val job = spark.readStream
+          .options(options.extraSourceOptions(tableName))
           .table(tableName)
           .writeStream
           .queryName(indexName)
-          .addIndexOptions(options)
+          .addSinkOptions(options)
           .foreachBatch { (batchDF: DataFrame, _: Long) =>
             batchRefresh(Some(batchDF))
           }
@@ -237,26 +238,26 @@ class FlintSpark(val spark: SparkSession) {
   // Using Scala implicit class to avoid breaking method chaining of Spark data frame fluent API
   private implicit class FlintDataStreamWriter(val dataStream: DataStreamWriter[Row]) {
 
-    def addIndexOptions(options: FlintSparkIndexOptions): DataStreamWriter[Row] = {
+    def addSinkOptions(options: FlintSparkIndexOptions): DataStreamWriter[Row] = {
       dataStream
         .addCheckpointLocation(options.checkpointLocation())
         .addRefreshInterval(options.refreshInterval())
+        .addOutputMode(options.outputMode())
+        .options(options.extraSinkOptions())
     }
 
     def addCheckpointLocation(checkpointLocation: Option[String]): DataStreamWriter[Row] = {
-      if (checkpointLocation.isDefined) {
-        dataStream.option("checkpointLocation", checkpointLocation.get)
-      } else {
-        dataStream
-      }
+      checkpointLocation.map(dataStream.option("checkpointLocation", _)).getOrElse(dataStream)
     }
 
     def addRefreshInterval(refreshInterval: Option[String]): DataStreamWriter[Row] = {
-      if (refreshInterval.isDefined) {
-        dataStream.trigger(Trigger.ProcessingTime(refreshInterval.get))
-      } else {
-        dataStream
-      }
+      refreshInterval
+        .map(interval => dataStream.trigger(Trigger.ProcessingTime(interval)))
+        .getOrElse(dataStream)
+    }
+
+    def addOutputMode(outputMode: Option[String]): DataStreamWriter[Row] = {
+      outputMode.map(dataStream.outputMode).getOrElse(dataStream)
     }
   }
 }

diff --git a/...-spark-integration/src/main/scala/org/opensearch/flint/spark/FlintSparkIndexOptions.scala b/...-spark-integration/src/main/scala/org/opensearch/flint/spark/FlintSparkIndexOptions.scala
@@ -5,7 +5,10 @@
 
 package org.opensearch.flint.spark
 
-import org.opensearch.flint.spark.FlintSparkIndexOptions.OptionName.{AUTO_REFRESH, CHECKPOINT_LOCATION, INDEX_SETTINGS, OptionName, REFRESH_INTERVAL}
+import org.json4s.{Formats, NoTypeHints}
+import org.json4s.native.JsonMethods._
+import org.json4s.native.Serialization
+import org.opensearch.flint.spark.FlintSparkIndexOptions.OptionName.{AUTO_REFRESH, CHECKPOINT_LOCATION, EXTRA_OPTIONS, INDEX_SETTINGS, OptionName, OUTPUT_MODE, REFRESH_INTERVAL, WATERMARK_DELAY}
 import org.opensearch.flint.spark.FlintSparkIndexOptions.validateOptionNames
 
 /**
@@ -16,6 +19,8 @@ import org.opensearch.flint.spark.FlintSparkIndexOptions.validateOptionNames
  */
 case class FlintSparkIndexOptions(options: Map[String, String]) {
 
+  implicit val formats: Formats = Serialization.formats(NoTypeHints)
+
   validateOptionNames(options)
 
   /**
@@ -42,6 +47,21 @@ case class FlintSparkIndexOptions(options: Map[String, String]) {
    */
   def checkpointLocation(): Option[String] = getOptionValue(CHECKPOINT_LOCATION)
 
+  /**
+   * How late the data can come and still be processed.
+   *
+   * @return
+   *   watermark delay time expression
+   */
+  def watermarkDelay(): Option[String] = getOptionValue(WATERMARK_DELAY)
+
+  /**
+   * The output mode that describes how data will be written to streaming sink.
+   * @return
+   *   output mode
+   */
+  def outputMode(): Option[String] = getOptionValue(OUTPUT_MODE)
+
   /**
    * The index settings for OpenSearch index created.
    *
@@ -50,6 +70,28 @@ case class FlintSparkIndexOptions(options: Map[String, String]) {
    */
   def indexSettings(): Option[String] = getOptionValue(INDEX_SETTINGS)
 
+  /**
+   * Extra streaming source options that can be simply passed to DataStreamReader or
+   * Relation.options
+   * @param source
+   *   source name (full table name)
+   * @return
+   *   extra source option map or empty map if not exist
+   */
+  def extraSourceOptions(source: String): Map[String, String] = {
+    parseExtraOptions(source)
+  }
+
+  /**
+   * Extra streaming sink options that can be simply passed to DataStreamWriter.options()
+   *
+   * @return
+   *   extra sink option map or empty map if not exist
+   */
+  def extraSinkOptions(): Map[String, String] = {
+    parseExtraOptions("sink")
+  }
+
   /**
    * @return
    *   all option values and fill default value if unspecified
@@ -67,6 +109,12 @@ case class FlintSparkIndexOptions(options: Map[String, String]) {
   private def getOptionValue(name: OptionName): Option[String] = {
     options.get(name.toString)
   }
+
+  private def parseExtraOptions(key: String): Map[String, String] = {
+    getOptionValue(EXTRA_OPTIONS)
+      .map(opt => (parse(opt) \ key).extract[Map[String, String]])
+      .getOrElse(Map.empty)
+  }
 }
 
 object FlintSparkIndexOptions {
@@ -84,7 +132,10 @@ object FlintSparkIndexOptions {
     val AUTO_REFRESH: OptionName.Value = Value("auto_refresh")
     val REFRESH_INTERVAL: OptionName.Value = Value("refresh_interval")
     val CHECKPOINT_LOCATION: OptionName.Value = Value("checkpoint_location")
+    val WATERMARK_DELAY: OptionName.Value = Value("watermark_delay")
+    val OUTPUT_MODE: OptionName.Value = Value("output_mode")
     val INDEX_SETTINGS: OptionName.Value = Value("index_settings")
+    val EXTRA_OPTIONS: OptionName.Value = Value("extra_options")
   }
 
   /**

diff --git a/...integration/src/main/scala/org/opensearch/flint/spark/mv/FlintSparkMaterializedView.scala b/...integration/src/main/scala/org/opensearch/flint/spark/mv/FlintSparkMaterializedView.scala
@@ -8,6 +8,7 @@ package org.opensearch.flint.spark.mv
 import java.util.Locale
 
 import scala.collection.JavaConverters.mapAsJavaMapConverter
+import scala.collection.convert.ImplicitConversions.`map AsScala`
 
 import org.opensearch.flint.core.metadata.FlintMetadata
 import org.opensearch.flint.spark.{FlintSpark, FlintSparkIndex, FlintSparkIndexBuilder, FlintSparkIndexOptions}
@@ -23,6 +24,7 @@ import org.apache.spark.sql.catalyst.expressions.Attribute
 import org.apache.spark.sql.catalyst.plans.logical.{Aggregate, EventTimeWatermark, LogicalPlan}
 import org.apache.spark.sql.catalyst.util.IntervalUtils
 import org.apache.spark.sql.flint.{logicalPlanToDataFrame, qualifyTableName}
+import org.apache.spark.sql.util.CaseInsensitiveStringMap
 
 /**
  * Flint materialized view in Spark.
@@ -44,9 +46,6 @@ case class FlintSparkMaterializedView(
     extends FlintSparkIndex
     with StreamingRefresh {
 
-  /** TODO: add it to index option */
-  private val watermarkDelay = "0 Minute"
-
   override val kind: String = MV_INDEX_TYPE
 
   override def name(): String = getFlintIndexName(mvName)
@@ -81,19 +80,33 @@ case class FlintSparkMaterializedView(
      *  2.Set isStreaming flag to true in Relation operator
      */
     val streamingPlan = batchPlan transform {
-      case WindowingAggregate(agg, timeCol) =>
-        agg.copy(child = watermark(timeCol, watermarkDelay, agg.child))
+      case WindowingAggregate(aggregate, timeCol) =>
+        aggregate.copy(child = watermark(timeCol, aggregate.child))
 
       case relation: UnresolvedRelation if !relation.isStreaming =>
-        relation.copy(isStreaming = true)
+        relation.copy(isStreaming = true, options = optionsWithExtra(spark, relation))
     }
     logicalPlanToDataFrame(spark, streamingPlan)
   }
 
-  private def watermark(timeCol: Attribute, delay: String, child: LogicalPlan) = {
+  private def watermark(timeCol: Attribute, child: LogicalPlan) = {
+    require(
+      options.watermarkDelay().isDefined,
+      "watermark delay is required for incremental refresh with aggregation")
+
+    val delay = options.watermarkDelay().get
     EventTimeWatermark(timeCol, IntervalUtils.fromIntervalString(delay), child)
   }
 
+  private def optionsWithExtra(
+      spark: SparkSession,
+      relation: UnresolvedRelation): CaseInsensitiveStringMap = {
+    val originalOptions = relation.options.asCaseSensitiveMap
+    val tableName = qualifyTableName(spark, relation.tableName)
+    val extraOptions = options.extraSourceOptions(tableName).asJava
+    new CaseInsensitiveStringMap((originalOptions ++ extraOptions).asJava)
+  }
+
   /**
    * Extractor that extract event time column out of Aggregate operator.
    */
@@ -107,7 +120,7 @@ case class FlintSparkMaterializedView(
 
       if (winFuncs.size != 1) {
         throw new IllegalStateException(
-          "A windowing function is required for streaming aggregation")
+          "A windowing function is required for incremental refresh with aggregation")
       }
 
       // Assume first aggregate item must be time column

diff --git a/...k-integration/src/test/scala/org/opensearch/flint/spark/FlintSparkIndexOptionsSuite.scala b/...k-integration/src/test/scala/org/opensearch/flint/spark/FlintSparkIndexOptionsSuite.scala
@@ -5,7 +5,7 @@
 
 package org.opensearch.flint.spark
 
-import org.opensearch.flint.spark.FlintSparkIndexOptions.OptionName.{AUTO_REFRESH, CHECKPOINT_LOCATION, INDEX_SETTINGS, REFRESH_INTERVAL}
+import org.opensearch.flint.spark.FlintSparkIndexOptions.OptionName._
 import org.scalatest.matchers.should.Matchers
 
 import org.apache.spark.FlintSuite
@@ -16,7 +16,10 @@ class FlintSparkIndexOptionsSuite extends FlintSuite with Matchers {
     AUTO_REFRESH.toString shouldBe "auto_refresh"
     REFRESH_INTERVAL.toString shouldBe "refresh_interval"
     CHECKPOINT_LOCATION.toString shouldBe "checkpoint_location"
+    WATERMARK_DELAY.toString shouldBe "watermark_delay"
+    OUTPUT_MODE.toString shouldBe "output_mode"
     INDEX_SETTINGS.toString shouldBe "index_settings"
+    EXTRA_OPTIONS.toString shouldBe "extra_options"
   }
 
   test("should return specified option value") {
@@ -25,12 +28,42 @@ class FlintSparkIndexOptionsSuite extends FlintSuite with Matchers {
         "auto_refresh" -> "true",
         "refresh_interval" -> "1 Minute",
         "checkpoint_location" -> "s3://test/",
-        "index_settings" -> """{"number_of_shards": 3}"""))
+        "watermark_delay" -> "30 Seconds",
+        "output_mode" -> "complete",
+        "index_settings" -> """{"number_of_shards": 3}""",
+        "extra_options" ->
+          """ {
+            |   "alb_logs": {
+            |     "opt1": "val1"
+            |   },
+            |   "sink": {
+            |     "opt2": "val2",
+            |     "opt3": "val3"
+            |   }
+            | }""".stripMargin))
 
     options.autoRefresh() shouldBe true
     options.refreshInterval() shouldBe Some("1 Minute")
     options.checkpointLocation() shouldBe Some("s3://test/")
+    options.watermarkDelay() shouldBe Some("30 Seconds")
+    options.outputMode() shouldBe Some("complete")
     options.indexSettings() shouldBe Some("""{"number_of_shards": 3}""")
+    options.extraSourceOptions("alb_logs") shouldBe Map("opt1" -> "val1")
+    options.extraSinkOptions() shouldBe Map("opt2" -> "val2", "opt3" -> "val3")
+  }
+
+  test("should return extra source option value and empty sink option values") {
+    val options = FlintSparkIndexOptions(
+      Map("extra_options" ->
+        """ {
+            |   "alb_logs": {
+            |     "opt1": "val1"
+            |   }
+            | }""".stripMargin))
+
+    options.extraSourceOptions("alb_logs") shouldBe Map("opt1" -> "val1")
+    options.extraSourceOptions("alb_logs_metrics") shouldBe empty
+    options.extraSinkOptions() shouldBe empty
   }
 
   test("should return default option value if unspecified") {
@@ -39,11 +72,15 @@ class FlintSparkIndexOptionsSuite extends FlintSuite with Matchers {
     options.autoRefresh() shouldBe false
     options.refreshInterval() shouldBe empty
     options.checkpointLocation() shouldBe empty
+    options.watermarkDelay() shouldBe empty
+    options.outputMode() shouldBe empty
     options.indexSettings() shouldBe empty
+    options.extraSourceOptions("alb_logs") shouldBe empty
+    options.extraSinkOptions() shouldBe empty
     options.optionsWithDefault should contain("auto_refresh" -> "false")
   }
 
-  test("should return default option value if unspecified with specified value") {
+  test("should return include unspecified option if it has default value") {
     val options = FlintSparkIndexOptions(Map("refresh_interval" -> "1 Minute"))
 
     options.optionsWithDefault shouldBe Map(