From e9a0591a91e55df8d47908b453ee7afce2a18c95 Mon Sep 17 00:00:00 2001
From: hui lai <laihui@selectdb.com>
Date: Fri, 29 Nov 2024 20:43:43 +0800
Subject: [PATCH] [improve](routine load) ensure load balance after scaling up
 BE nodes (#44693)

We scaling up from three BE nodes to five BE nodes, but from the
monitoring perspective, only the previous three nodes have written
traffic.

![image](https://github.com/user-attachments/assets/947fe8a9-745d-4c37-93d4-15b0e27f12fc)

This pr aims to ensure load balance after scaling up BE nodes.
---
 .../org/apache/doris/load/routineload/RoutineLoadManager.java    | 1 +
 1 file changed, 1 insertion(+)

diff --git a/fe/fe-core/src/main/java/org/apache/doris/load/routineload/RoutineLoadManager.java b/fe/fe-core/src/main/java/org/apache/doris/load/routineload/RoutineLoadManager.java
index 6121f34035ce06..c4cb6119e94525 100644
--- a/fe/fe-core/src/main/java/org/apache/doris/load/routineload/RoutineLoadManager.java
+++ b/fe/fe-core/src/main/java/org/apache/doris/load/routineload/RoutineLoadManager.java
@@ -486,6 +486,7 @@ public long getAvailableBeForTask(long jobId, long previousBeId) throws LoadExce
         // check if be has idle slot
         readLock();
         try {
+            updateBeIdToMaxConcurrentTasks();
             Map<Long, Integer> beIdToConcurrentTasks = getBeCurrentTasksNumMap();
             int previousBeIdleTaskNum = 0;