I managed to reproduce the issue on the RHEL8 worker using:

* terminal 1: ./python -u -m test --fail-env-changed test_concurrent_futures -v -m ProcessPoolSpawnProcessPoolExecutorTest -F -j 15
* terminal 2: ./python -m test -j4 -u all -F -r

I can still reproduce the issue with PR 17641 fix, so it's not enough.

I also applied PR 17640 and 2 extra changes:

diff --git a/Lib/multiprocessing/managers.py b/Lib/multiprocessing/managers.py
index 1f9c2daa25..dbbf84b3d2 100644
--- a/Lib/multiprocessing/managers.py
+++ b/Lib/multiprocessing/managers.py
@@ -661,7 +661,7 @@ class BaseManager(object):
             except Exception:
                 pass
 
-            process.join(timeout=1.0)
+            process.join(timeout=0.1)
             if process.is_alive():
                 util.info('manager still alive')
                 if hasattr(process, 'terminate'):
diff --git a/Lib/test/support/__init__.py b/Lib/test/support/__init__.py
index 215bab8131..110c7f945e 100644
--- a/Lib/test/support/__init__.py
+++ b/Lib/test/support/__init__.py
@@ -2399,6 +2399,7 @@ def reap_children():
 
     # Reap all our dead child processes so we don't leave zombies around.
     # These hog resources and might be causing some of the buildbots to die.
+    time.sleep(0.5)
     while True:
         try:
             # Read the exit status of any child process which already completed