This is an automated email from the ASF dual-hosted git repository.

rexxiong pushed a commit to branch branch-0.3
in repository https://gitbox.apache.org/repos/asf/incubator-celeborn.git


The following commit(s) were added to refs/heads/branch-0.3 by this push:
     new e5bd8073b [CELEBORN-1026] Optimize registerShuffle fallback log
e5bd8073b is described below

commit e5bd8073b0b2e8b6a1177798ce72c6e5c5b0fe66
Author: sychen <[email protected]>
AuthorDate: Wed Oct 11 20:25:44 2023 +0800

    [CELEBORN-1026] Optimize registerShuffle fallback log
    
    ### What changes were proposed in this pull request?
    
    ### Why are the changes needed?
    According to https://github.com/apache/incubator-celeborn/pull/1955 , when 
celeborn is not available, `spark.dynamicAllocation.enabled=true` and  
`spark.shuffle.service. enabled=true`, shuffle data should not be lost.
    
    ### Does this PR introduce _any_ user-facing change?
    
    ### How was this patch tested?
    
    Closes #1963 from cxzl25/CELEBORN-1026.
    
    Authored-by: sychen <[email protected]>
    Signed-off-by: Shuang <[email protected]>
    (cherry picked from commit 62ba44d8daf39ab13726a3579e0adc2e47fa226f)
    Signed-off-by: Shuang <[email protected]>
---
 .../java/org/apache/spark/shuffle/celeborn/SparkShuffleManager.java    | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git 
a/client-spark/spark-3/src/main/java/org/apache/spark/shuffle/celeborn/SparkShuffleManager.java
 
b/client-spark/spark-3/src/main/java/org/apache/spark/shuffle/celeborn/SparkShuffleManager.java
index eb530555f..e93a6b40a 100644
--- 
a/client-spark/spark-3/src/main/java/org/apache/spark/shuffle/celeborn/SparkShuffleManager.java
+++ 
b/client-spark/spark-3/src/main/java/org/apache/spark/shuffle/celeborn/SparkShuffleManager.java
@@ -127,7 +127,8 @@ public class SparkShuffleManager implements ShuffleManager {
 
     if (fallbackPolicyRunner.applyAllFallbackPolicy(
         lifecycleManager, dependency.partitioner().numPartitions())) {
-      if (conf.getBoolean("spark.dynamicAllocation.enabled", false)) {
+      if (conf.getBoolean("spark.dynamicAllocation.enabled", false)
+          && !conf.getBoolean("spark.shuffle.service.enabled", false)) {
         logger.error(
             "DRA is enabled but we fallback to vanilla Spark 
SortShuffleManager for "
                 + "shuffle: {} due to fallback policy. It may cause block can 
not found when reducer "

Reply via email to