[GitHub] [arrow-ballista] mingmwang commented on a diff in pull request #261: Task level retry and Stage level retry

[GitBox <gi...@apache.org>]

mingmwang commented on code in PR #261:
URL: https://github.com/apache/arrow-ballista/pull/261#discussion_r985173900


##########
ballista/rust/core/src/error.rs:
##########
@@ -182,13 +200,78 @@ impl Display for BallistaError {
             // }
             BallistaError::TonicError(desc) => write!(f, "Tonic error: {}", desc),
             BallistaError::GrpcError(desc) => write!(f, "Grpc error: {}", desc),
+            BallistaError::GrpcConnectionError(desc) => {
+                write!(f, "Grpc connection error: {}", desc)
+            }
             BallistaError::Internal(desc) => {
                 write!(f, "Internal Ballista error: {}", desc)
             }
             BallistaError::TokioError(desc) => write!(f, "Tokio join error: {}", desc),
+            BallistaError::GrpcActionError(desc) => {
+                write!(f, "Grpc Execute Action error: {}", desc)
+            }
+            BallistaError::FetchFailed(executor_id, map_stage, map_partition, desc) => {
+                write!(
+                    f,
+                    "Shuffle fetch partition error from Executor {}, map_stage {}, \
+                map_partition {}, error desc: {}",
+                    executor_id, map_stage, map_partition, desc
+                )
+            }
             BallistaError::Cancelled => write!(f, "Task cancelled"),
         }
     }
 }
 
+impl From<BallistaError> for FailedTask {
+    fn from(e: BallistaError) -> Self {
+        match e {
+            BallistaError::FetchFailed(
+                executor_id,
+                map_stage_id,
+                map_partition_id,
+                desc,
+            ) => {
+                FailedTask {
+                    error: desc,
+                    // fetch partition error is considered to be non-retryable
+                    retryable: false,
+                    count_to_failures: false,
+                    failed_reason: Some(FailedReason::FetchPartitionError(
+                        FetchPartitionError {
+                            executor_id,
+                            map_stage_id: map_stage_id as u32,
+                            map_partition_id: map_partition_id as u32,
+                        },
+                    )),
+                }
+            }
+            BallistaError::IoError(io) => {
+                FailedTask {
+                    error: format!("Task failed due to Ballista IO error: {:?}", io),
+                    // IO error is considered to be temporary and retryable
+                    retryable: true,
+                    count_to_failures: true,
+                    failed_reason: Some(FailedReason::IoError(IoError {})),
+                }
+            }
+            BallistaError::DataFusionError(DataFusionError::IoError(io)) => {
+                FailedTask {
+                    error: format!("Task failed due to DataFusion IO error: {:?}", io),
+                    // IO error is considered to be temporary and retryable
+                    retryable: true,
+                    count_to_failures: true,

Review Comment:
   Yes, have both `retyrable` `and count_to_failures` is to support the case that we might have some specific error that we want it to retry forever until it is successful.



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org