[GitHub] [hadoop] hfutatzhanghb commented on a diff in pull request #5778: HDFS-17060. BlockPlacementPolicyDefault#chooseReplicaToDelete should consider datanode load.

["hfutatzhanghb (via GitHub)" <gi...@apache.org>]

hfutatzhanghb commented on code in PR #5778:
URL: https://github.com/apache/hadoop/pull/5778#discussion_r1241488678


##########
hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/blockmanagement/BlockPlacementPolicyDefault.java:
##########
@@ -1236,11 +1245,17 @@ public DatanodeStorageInfo chooseReplicaToDelete(
         minSpace = free;
         minSpaceStorage = storage;
       }
+      if (considerLoad2chooseReplicaDeleting && nodeLoad < minLoad) {
+        minLoad = nodeLoad;
+        minLoadStorage = storage;
+      }
     }
 
     final DatanodeStorageInfo storage;
     if (oldestHeartbeatStorage != null) {
       storage = oldestHeartbeatStorage;
+    } else if (minLoadStorage != null) {

Review Comment:
   @Hexiaoqiao Sir, thanks for your reply.  If we open the configuration mentioned in this PR,  we should consider load first and we keep running Balancer background.  And I also have thinked some improvements about the changes:
   1. should we consider load use `load.factor` like `dfs.namenode.redundancy.considerLoad.factor`?
   2. should we add another configuration to judge the difference between two storage's remaining in an acceptable range?
   
   Looking forward to your reply~ Thanks sir.



-- 
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: common-issues-unsubscribe@hadoop.apache.org

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


---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: common-issues-help@hadoop.apache.org