diff --git a/CHANGELOG.md b/CHANGELOG.md
index 81f64b333c..5f7949feed 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -25,6 +25,7 @@ CHANGELOG
 - Fail `pcluster build-image` early when the downloaded cookbook version does not match the ParallelCluster CLI version.
 - Fix login nodes not mounting `/opt/parallelcluster/shared` when EFS is used as the internal shared storage type.
 - Fix an issue where compute nodes are incorrectly replaced when launching a large number of nodes due to eventual consistency.
+- Fix an issue where starting the compute fleet may not reliably recover the cluster from protected mode.
 
 **DEPRECATIONS**
 - Amazon Linux 2 is no longer supported.
diff --git a/tests/integration-tests/tests/schedulers/test_slurm.py b/tests/integration-tests/tests/schedulers/test_slurm.py
index ad57043214..7a424a5853 100644
--- a/tests/integration-tests/tests/schedulers/test_slurm.py
+++ b/tests/integration-tests/tests/schedulers/test_slurm.py
@@ -379,13 +379,6 @@ def test_slurm_custom_partitions(
         assert_that(scheduler_commands.get_partition_state(partition=partition)).is_equal_to(expected_state)
 
     logging.info("Checking pcluster start...")
-    # Restore the protected failure count to its default before starting the fleet. The failing job left dynamic nodes
-    # still powering up when the fleet was stopped; with a short clustermgtd cleanup window they can survive stop/start
-    # with a stale nodeaddr. On the first poll after start, clustermgtd re-counts them as bootstrap failures and, with
-    # the lowered count of 2, re-enters protected mode, flipping the pcluster-managed partitions back to INACTIVE and
-    # making this test intermittently fail. The default count is high enough that the few leftover nodes cannot reach
-    # the threshold in a single poll.
-    set_protected_failure_count(remote_command_executor, 10)
     for partition in custom_partitions:
         scheduler_commands.set_partition_state(partition, "INACTIVE")
     cluster.start()