[improvement] 2pc commit or abort retry and interval configurable.

apache · Sep 7, 2023 · d7ffc97 · d7ffc97
1 parent 11f4976
commit d7ffc97
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 6 deletions.
diff --git a/spark-doris-connector/src/main/java/org/apache/doris/spark/cfg/ConfigurationOptions.java b/spark-doris-connector/src/main/java/org/apache/doris/spark/cfg/ConfigurationOptions.java
@@ -106,4 +106,16 @@ public interface ConfigurationOptions {
     String DORIS_SINK_STREAMING_PASSTHROUGH = "doris.sink.streaming.passthrough";
     boolean DORIS_SINK_STREAMING_PASSTHROUGH_DEFAULT = false;
 
+    /**
+     * txnId commit or abort interval
+     */
+    String DORIS_SINK_TXN_INTERVAL_MS = "doris.sink.txn.interval.ms";
+    int DORIS_SINK_TXN_INTERVAL_MS_DEFAULT = 1000;
+
+    /**
+     * txnId commit or abort retry times
+     */
+    String DORIS_SINK_TXN_RETRIES = "doris.sink.txn.retries";
+    int DORIS_SINK_TXN_RETRIES_DEFAULT = 3;
+
 }
diff --git a/...s-connector/src/main/scala/org/apache/doris/spark/listener/DorisTransactionListener.scala b/...s-connector/src/main/scala/org/apache/doris/spark/listener/DorisTransactionListener.scala
@@ -28,7 +28,7 @@ import scala.collection.JavaConverters._
 import scala.collection.mutable
 import scala.util.{Failure, Success}
 
-class DorisTransactionListener(preCommittedTxnAcc: CollectionAccumulator[Int], dorisStreamLoad: DorisStreamLoad)
+class DorisTransactionListener(preCommittedTxnAcc: CollectionAccumulator[Int], dorisStreamLoad: DorisStreamLoad, sinkTnxIntervalMs: Int, sinkTxnRetries: Int)
   extends SparkListener {
 
   val logger: Logger = LoggerFactory.getLogger(classOf[DorisTransactionListener])
@@ -45,7 +45,7 @@ class DorisTransactionListener(preCommittedTxnAcc: CollectionAccumulator[Int], d
         }
         logger.info("job run succeed, start committing transactions")
         txnIds.foreach(txnId =>
-          Utils.retry(3, Duration.ofSeconds(1), logger) {
+          Utils.retry(sinkTxnRetries, Duration.ofMillis(sinkTnxIntervalMs), logger) {
             dorisStreamLoad.commit(txnId)
           } match {
             case Success(_) =>
@@ -66,7 +66,7 @@ class DorisTransactionListener(preCommittedTxnAcc: CollectionAccumulator[Int], d
         }
         logger.info("job run failed, start aborting transactions")
         txnIds.foreach(txnId =>
-          Utils.retry(3, Duration.ofSeconds(1), logger) {
+          Utils.retry(sinkTxnRetries, Duration.ofMillis(sinkTnxIntervalMs), logger) {
             dorisStreamLoad.abort(txnId)
           } match {
             case Success(_) =>

diff --git a/spark-doris-connector/src/main/scala/org/apache/doris/spark/writer/DorisWriter.scala b/spark-doris-connector/src/main/scala/org/apache/doris/spark/writer/DorisWriter.scala
@@ -51,6 +51,10 @@ class DorisWriter(settings: SparkSettings) extends Serializable {
 
   private val enable2PC: Boolean = settings.getBooleanProperty(ConfigurationOptions.DORIS_SINK_ENABLE_2PC,
     ConfigurationOptions.DORIS_SINK_ENABLE_2PC_DEFAULT);
+  private val sinkTxnIntervalMs: Int = settings.getIntegerProperty(ConfigurationOptions.DORIS_SINK_TXN_INTERVAL_MS,
+    ConfigurationOptions.DORIS_SINK_TXN_INTERVAL_MS_DEFAULT)
+  private val sinkTxnRetries: Integer = settings.getIntegerProperty(ConfigurationOptions.DORIS_SINK_TXN_RETRIES,
+    ConfigurationOptions.DORIS_SINK_TXN_RETRIES_DEFAULT)
 
   private val dorisStreamLoader: DorisStreamLoad = CachedDorisStreamLoadClient.getOrCreate(settings)
 
@@ -59,7 +63,7 @@ class DorisWriter(settings: SparkSettings) extends Serializable {
     val sc = dataFrame.sqlContext.sparkContext
     val preCommittedTxnAcc = sc.collectionAccumulator[Int]("preCommittedTxnAcc")
     if (enable2PC) {
-      sc.addSparkListener(new DorisTransactionListener(preCommittedTxnAcc, dorisStreamLoader))
+      sc.addSparkListener(new DorisTransactionListener(preCommittedTxnAcc, dorisStreamLoader, sinkTxnIntervalMs, sinkTxnRetries))
     }
 
     var resultRdd = dataFrame.rdd
@@ -98,7 +102,7 @@ class DorisWriter(settings: SparkSettings) extends Serializable {
     val sc = dataFrame.sqlContext.sparkContext
     val preCommittedTxnAcc = sc.collectionAccumulator[Int]("preCommittedTxnAcc")
     if (enable2PC) {
-      sc.addSparkListener(new DorisTransactionListener(preCommittedTxnAcc, dorisStreamLoader))
+      sc.addSparkListener(new DorisTransactionListener(preCommittedTxnAcc, dorisStreamLoader, sinkTxnIntervalMs, sinkTxnRetries))
     }
 
     var resultRdd = dataFrame.queryExecution.toRdd
@@ -153,7 +157,7 @@ class DorisWriter(settings: SparkSettings) extends Serializable {
     }
     val abortFailedTxnIds = mutable.Buffer[Int]()
     acc.value.asScala.foreach(txnId => {
-      Utils.retry[Unit, Exception](3, Duration.ofSeconds(1), logger) {
+      Utils.retry[Unit, Exception](sinkTxnRetries, Duration.ofMillis(sinkTxnIntervalMs), logger) {
         dorisStreamLoader.abort(txnId)
       } match {
         case Success(_) =>