Retry Refinements

- Update readme, including version of update
- Add 2 additional plausible retriable 400 issue's

README.md

@@ -125,7 +125,7 @@ Changing this number changes both the number of concurrent connections and the u
 
 ## Retry behavior
 
-This plugin uses exponential backoff when 50X errors from DynamoDB would occur. This includes network glitches and provisioned throughput exceptions (request rate is too high).
+This plugin uses exponential backoff when plausible and retriable errors from DynamoDB occur. This includes network glitches (50x; since Pekko 1.1.0) and throughput exceptions (400; extended in Pekko 1.1.0).
 
 The backoff strategy is very simple. 
 - There are a maximum of 10 retries.

src/main/scala/org/apache/pekko/persistence/dynamodb/journal/DynamoDBHelper.scala

@@ -38,13 +38,23 @@ private class RetryStateHolder(var retries: Int = 10, var backoff: FiniteDuratio
 object DynamoRetriableException {
   def unapply(ex: AmazonServiceException) = {
     ex match {
-      case _: ProvisionedThroughputExceededException =>
-        Some(ex)
+      // 50x network glitches
       case _: InternalServerErrorException =>
         Some(ex)
       case ase if ase.getStatusCode >= 502 && ase.getStatusCode <= 504 =>
         // retry on more common server errors
         Some(ex)
+
+      // 400 throughput issues
+      case _: ProvisionedThroughputExceededException =>
+        Some(ex)
+      case _: RequestLimitExceededException =>
+        // rate of on-demand requests exceeds the allowed account throughput
+        // and the table cannot be scaled further
+        Some(ex)
+      case ase if ase.getErrorCode == "ThrottlingException" =>
+        // // rate of AWS requests exceeds the allowed throughput
+        Some(ex)
       case _ =>
         None
     }