Fix deadlock

clearml/backend_interface/logger.py

@@ -1,10 +1,10 @@
 import logging
 import sys
-import threading
 from time import time
 
 from ..binding.frameworks import _patched_call  # noqa
 from ..config import running_remotely, config, DEBUG_SIMULATE_REMOTE_TASK
+from ..utilities.process.mp import ForkSafeRLock
 
 
 class StdStreamPatch(object):
@@ -176,8 +176,8 @@ class PrintPatchLogger(object):
     Used for capturing and logging stdin and stderr when running in development mode pseudo worker.
     """
     patched = False
-    lock = threading.Lock()
-    recursion_protect_lock = threading.RLock()
+    lock = ForkSafeRLock()
+    recursion_protect_lock = ForkSafeRLock()
     cr_flush_period = None
 
     def __init__(self, stream, logger=None, level=logging.INFO, load_config_defaults=True):

clearml/binding/environ_bind.py

@@ -76,8 +76,8 @@ class SimpleQueueWrapper(object):
             def _patched_put(*a_args, **a_kwargs):
                 # make sure we flush everything, because after we push the result we will get terminated
                 try:
-                    task = self.__current_task
-                    task.flush(wait_for_uploads=True)
+                    if self.__current_task and self.__current_task.is_main_task():
+                        self.__current_task.flush(wait_for_uploads=True)
                 except:  # noqa
                     pass
                 return getattr(self.__simple_queue, "put")(*a_args, **a_kwargs)
@@ -152,6 +152,7 @@ class PatchOsFork(object):
             task = None
 
         # check if this is Process Pool function
+        patched_worker = False
         if hasattr(self, "_target"):
             # Now we have to patch Pool, because pool terminates subprocess directly after
             # the return value of the pool worker function is pushed into the queue,
@@ -159,15 +160,22 @@ class PatchOsFork(object):
             try:
                 if self._target == pool.worker:  # noqa
                     self._target = partial(PatchOsFork._patched_pool_worker, pool.worker)  # noqa
+                    patched_worker = True
             except:  # noqa
                 pass
 
         try:
             return PatchOsFork._original_process_run(self, *args, **kwargs)
         finally:
-            # force creating a Task
-            try:
             if task:
+                try:
+                    if patched_worker:
+                        # remove at exit hooks, we will deadlock when the
+                        # main Pool manager will terminate this process, and it will...
+                        # noinspection PyProtectedMember
+                        task._at_exit_called = True
+                    else:
+                        # terminate the current Task
                         # noinspection PyProtectedMember
                         task._at_exit()
                 except:  # noqa

clearml/utilities/process/mp.py

@@ -142,6 +142,10 @@ class ForkSafeRLock(_ForkSafeThreadSyncObject):
         # Do whatever cleanup.
         self.release()
 
+    def _is_owned(self):
+        self._create()
+        return self._sync._is_owned()  # noqa
+
 
 class ForkSemaphore(_ForkSafeThreadSyncObject):
     def __init__(self, value=1):