Cache tasks as they are queued instead of en masse (#34752)

* Cache tasks as they are queued instead of en masse

This also moves the task caching from the PlayIterator to the
StrategyBase class, where it makes more sense (and makes it easier
to not have to change the strategy class methods leading to an API
change).

Fixes #31673

* Cleaning up unit tests due to 502ca780

2 files changed, 23 insertions, 41 deletions

diff --git a/lib/ansible/executor/play_iterator.py b/lib/ansible/executor/play_iterator.py
index 99b7285182..e9f6a0c47b 100644
--- a/lib/ansible/executor/play_iterator.py
+++ b/lib/ansible/executor/play_iterator.py
@@ -153,8 +153,6 @@ class PlayIterator:
         self._blocks = []
         self._variable_manager = variable_manager
 
-        self._task_uuid_cache = dict()
-
         # Default options to gather
         gather_subset = play_context.gather_subset
         gather_timeout = play_context.gather_timeout
@@ -242,16 +240,10 @@ class PlayIterator:
         return self._host_states[host.name].copy()
 
     def cache_block_tasks(self, block):
-        def _cache_portion(p):
-            for t in p:
-                if isinstance(t, Block):
-                    self.cache_block_tasks(t)
-                elif t._uuid not in self._task_uuid_cache:
-                    self._task_uuid_cache[t._uuid] = t
-
-        for portion in (block.block, block.rescue, block.always):
-            if portion is not None:
-                _cache_portion(portion)
+        # now a noop, we've changed the way we do caching and finding of
+        # original task entries, but just in case any 3rd party strategies
+        # are using this we're leaving it here for now
+        return
 
     def get_next_task_for_host(self, host, peek=False):
 
@@ -520,19 +512,8 @@ class PlayIterator:
         return self._check_failed_state(s)
 
     def get_original_task(self, host, task):
-        '''
-        Finds the task in the task list which matches the UUID of the given task.
-        The executor engine serializes/deserializes objects as they are passed through
-        the different processes, and not all data structures are preserved. This method
-        allows us to find the original task passed into the executor engine.
-        '''
-
-        if isinstance(task, Task):
-            the_uuid = task._uuid
-        else:
-            the_uuid = task
-
-        return self._task_uuid_cache.get(the_uuid, None)
+        # now a noop because we've changed the way we do caching
+        return (None, None)
 
     def _insert_tasks_into_state(self, state, task_list):
         # if we've failed at all, or if the task list is empty, just return the current state
@@ -569,6 +550,4 @@ class PlayIterator:
         return state
 
     def add_tasks(self, host, task_list):
-        for b in task_list:
-            self.cache_block_tasks(b)
         self._host_states[host.name] = self._insert_tasks_into_state(self.get_host_state(host), task_list)
diff --git a/lib/ansible/plugins/strategy/__init__.py b/lib/ansible/plugins/strategy/__init__.py
index 4eb08901b8..7b5d3348a8 100644
--- a/lib/ansible/plugins/strategy/__init__.py
+++ b/lib/ansible/plugins/strategy/__init__.py
@@ -120,9 +120,9 @@ def debug_closure(func):
         for result in results:
             task = result._task
             host = result._host
-            _queue_task_args = self._queue_task_args.pop('%s%s' % (host.name, task._uuid))
-            task_vars = _queue_task_args['task_vars']
-            play_context = _queue_task_args['play_context']
+            _queued_task_args = self._queued_task_cache.pop((host.name, task._uuid), None)
+            task_vars = _queued_task_args['task_vars']
+            play_context = _queued_task_args['play_context']
             # Try to grab the previous host state, if it doesn't exist use get_host_state to generate an empty state
             try:
                 prev_host_state = prev_host_states[host.name]
@@ -179,7 +179,11 @@ class StrategyBase:
         self._final_q = tqm._final_q
         self._step = getattr(tqm._options, 'step', False)
         self._diff = getattr(tqm._options, 'diff', False)
-        self._queue_task_args = {}
+
+        # the task cache is a dictionary of tuples of (host.name, task._uuid)
+        # used to find the original task object of in-flight tasks and to store
+        # the task args/vars and play context info used to queue the task.
+        self._queued_task_cache = {}
 
         # Backwards compat: self._display isn't really needed, just import the global display and use that.
         self._display = display
@@ -270,13 +274,6 @@ class StrategyBase:
     def _queue_task(self, host, task, task_vars, play_context):
         ''' handles queueing the task up to be sent to a worker '''
 
-        self._queue_task_args['%s%s' % (host.name, task._uuid)] = {
-            'host': host,
-            'task': task,
-            'task_vars': task_vars,
-            'play_context': play_context
-        }
-
         display.debug("entering _queue_task() for %s/%s" % (host.name, task.action))
 
         # Add a write lock for tasks.
@@ -306,6 +303,13 @@ class StrategyBase:
             while True:
                 (worker_prc, rslt_q) = self._workers[self._cur_worker]
                 if worker_prc is None or not worker_prc.is_alive():
+                    self._queued_task_cache[(host.name, task._uuid)] = {
+                        'host': host,
+                        'task': task,
+                        'task_vars': task_vars,
+                        'play_context': play_context
+                    }
+
                     worker_prc = WorkerProcess(self._final_q, task_vars, host, task, play_context, self._loader, self._variable_manager, shared_loader_obj)
                     self._workers[self._cur_worker][0] = worker_prc
                     worker_prc.start()
@@ -425,7 +429,8 @@ class StrategyBase:
 
             # get the original host and task. We then assign them to the TaskResult for use in callbacks/etc.
             original_host = get_original_host(task_result._host)
-            found_task = iterator.get_original_task(original_host, task_result._task)
+            queue_cache_entry = (original_host.name, task_result._task)
+            found_task = self._queued_task_cache.get(queue_cache_entry)['task']
             original_task = found_task.copy(exclude_parent=True, exclude_tasks=True)
             original_task._parent = found_task._parent
             original_task.from_attrs(task_result._task_fields)
@@ -854,8 +859,6 @@ class StrategyBase:
         host_results = []
         for host in notified_hosts:
             if not handler.has_triggered(host) and (not iterator.is_failed(host) or play_context.force_handlers):
-                if handler._uuid not in iterator._task_uuid_cache:
-                    iterator._task_uuid_cache[handler._uuid] = handler
                 task_vars = self._variable_manager.get_vars(play=iterator._play, host=host, task=handler)
                 self.add_tqm_variables(task_vars, play=iterator._play)
                 self._queue_task(host, handler, task_vars, play_context)