diff --git a/src/DIRAC/DataManagementSystem/Agent/FTS3Agent.py b/src/DIRAC/DataManagementSystem/Agent/FTS3Agent.py
index 4151a3fffb8..4e009158d43 100644
--- a/src/DIRAC/DataManagementSystem/Agent/FTS3Agent.py
+++ b/src/DIRAC/DataManagementSystem/Agent/FTS3Agent.py
@@ -106,6 +106,8 @@ def __readConf(self):
         # lifetime of the proxy we download to delegate to FTS
         self.proxyLifetime = self.am_getOption("ProxyLifetime", PROXY_LIFETIME)
 
+        self.jobMonitoringBatchSize = self.am_getOption("JobMonitoringBatchSize", JOB_MONITORING_BATCH_SIZE)
+
         return S_OK()
 
     def initialize(self):
@@ -318,7 +320,7 @@ def monitorJobsLoop(self):
             log.info("Getting next batch of jobs to monitor", f"{loopId}/{nbOfLoops}")
             # get jobs from DB
             res = self.fts3db.getActiveJobs(
-                limit=JOB_MONITORING_BATCH_SIZE, lastMonitor=lastMonitor, jobAssignmentTag=self.assignmentTag
+                limit=self.jobMonitoringBatchSize, lastMonitor=lastMonitor, jobAssignmentTag=self.assignmentTag
             )
 
             if not res["OK"]:
@@ -353,7 +355,7 @@ def monitorJobsLoop(self):
 
             # If we got less to monitor than what we asked,
             # stop looping
-            if len(activeJobs) < JOB_MONITORING_BATCH_SIZE:
+            if len(activeJobs) < self.jobMonitoringBatchSize:
                 break
         # Commit records after each loop
         self.dataOpSender.concludeSending()
diff --git a/src/DIRAC/DataManagementSystem/ConfigTemplate.cfg b/src/DIRAC/DataManagementSystem/ConfigTemplate.cfg
index 406f9d90bc7..c4b9636dab2 100644
--- a/src/DIRAC/DataManagementSystem/ConfigTemplate.cfg
+++ b/src/DIRAC/DataManagementSystem/ConfigTemplate.cfg
@@ -143,6 +143,12 @@ Agents
     OperationBulkSize = 20
     # How many Job we will monitor in one loop
     JobBulkSize = 20
+    # split jobBulkSize in several chunks
+    # Bigger numbers (like 100) are efficient when there's a single agent
+    # When there are multiple agents, it may slow down the overall because
+    # of lock and race conditions
+    # (This number should of course be smaller or equal than JobBulkSize)
+    JobMonitoringBatchSize = 20
     # Max number of files to go in a single job
     MaxFilesPerJob = 100
     # Max number of attempt per file