diff --git a/libs/utils/executor.py b/libs/utils/executor.py
index 7c7060b22fed9703c23816b4fd1155d0055d7a78..883c07470ef62e439a0dd446a6f0e217a121418a 100644
--- a/libs/utils/executor.py
+++ b/libs/utils/executor.py
@@ -166,6 +166,13 @@ class Executor():
             Number of iterations for each workload/conf combination. Default
             is 1.
     :type experiments_conf: dict
+
+    :ivar experiments: After calling `meth`:run:, the list of
+                       :class:`Experiment` s that were run
+
+    :ivar iterations: The number of iterations run for each wload/conf pair
+                       (i.e. ``experiments_conf['iterations']``.
+
     """
 
     critical_tasks = {
@@ -215,9 +222,9 @@ class Executor():
         self.te = test_env
         self.target = self.te.target
 
-        self._iterations = self._experiments_conf.get('iterations', 1)
+        self.iterations = self._experiments_conf.get('iterations', 1)
         # Compute total number of experiments
-        self._exp_count = self._iterations \
+        self._exp_count = self.iterations \
                 * len(self._experiments_conf['wloads']) \
                 * len(self._experiments_conf['confs'])
 
@@ -233,7 +240,7 @@ class Executor():
 
         self._log.info('   %3d workloads (%d iterations each)',
                        len(self._experiments_conf['wloads']),
-                       self._iterations)
+                       self.iterations)
         wload_confs = ', '.join(self._experiments_conf['wloads'])
         self._log.info('      %s', wload_confs)
 
@@ -261,7 +268,7 @@ class Executor():
             for wl_idx in self._experiments_conf['wloads']:
                 # TEST: configuration
                 wload, test_dir = self._wload_init(tc, wl_idx)
-                for itr_idx in range(1, self._iterations + 1):
+                for itr_idx in range(1, self.iterations + 1):
                     exp = Experiment(
                         wload_name=wl_idx,
                         wload=wload,
@@ -663,7 +670,7 @@ class Executor():
         self._print_title('Experiment {}/{}, [{}:{}] {}/{}'\
                 .format(exp_idx, self._exp_count,
                         tc_idx, experiment.wload_name,
-                        experiment.iteration, self._iterations))
+                        experiment.iteration, self.iterations))
 
         # Setup local results folder
         self._log.debug('out_dir set to [%s]', experiment.out_dir)
diff --git a/libs/utils/test.py b/libs/utils/test.py
index b6062168404c1baa5162ad240901fce6d6f56ad9..d3b96ec45a546406c926ac6ea22922192a58db88 100644
--- a/libs/utils/test.py
+++ b/libs/utils/test.py
@@ -56,6 +56,9 @@ class LisaTest(unittest.TestCase):
     experiments_conf = None
     """Override this with a dictionary or JSON path to configure the Executor"""
 
+    permitted_fail_pct = 0
+    """The percentage of iterations of each test that may be permitted to fail"""
+
     @classmethod
     def _getTestConf(cls):
         if cls.test_conf is None:
@@ -81,9 +84,9 @@ class LisaTest(unittest.TestCase):
         """
         Set up logging and trigger running experiments
         """
-        cls.logger = logging.getLogger('LisaTest')
+        cls._log = logging.getLogger('LisaTest')
 
-        cls.logger.info('Setup tests execution engine...')
+        cls._log.info('Setup tests execution engine...')
         test_env = TestEnv(test_conf=cls._getTestConf())
 
         experiments_conf = cls._getExperimentsConf(test_env)
@@ -100,7 +103,7 @@ class LisaTest(unittest.TestCase):
         # Execute pre-experiments code defined by the test
         cls._experimentsInit()
 
-        cls.logger.info('Experiments execution...')
+        cls._log.info('Experiments execution...')
         cls.executor.run()
 
         cls.experiments = cls.executor.experiments
@@ -213,16 +216,33 @@ def experiment_test(wrapped_test, instance, args, kwargs):
     The method will be passed the experiment object and a list of the names of
     tasks that were run as the experiment's workload.
     """
+    failures = {}
     for experiment in instance.executor.experiments:
         tasks = experiment.wload.tasks.keys()
         try:
             wrapped_test(experiment, tasks, *args, **kwargs)
         except AssertionError as e:
             trace_relpath = os.path.join(experiment.out_dir, "trace.dat")
-            add_msg = "\n\tCheck trace file: " + os.path.abspath(trace_relpath)
-            orig_msg = e.args[0] if len(e.args) else ""
-            e.args = (orig_msg + add_msg,) + e.args[1:]
-            raise
+            add_msg = "Check trace file: " + os.path.abspath(trace_relpath)
+            msg = str(e) + "\n\t" +  add_msg
+
+            test_key = (experiment.wload_name, experiment.conf['tag'])
+            failures[test_key] = failures.get(test_key, []) + [msg]
+
+    for fails in failures.itervalues():
+        iterations = instance.executor.iterations
+        fail_pct = 100. * len(fails) / iterations
+
+        msg = "{} failures from {} iteration(s):\n{}".format(
+            len(fails), iterations, '\n'.join(fails))
+        if fail_pct > instance.permitted_fail_pct:
+            raise AssertionError(msg)
+        else:
+            instance._log.warning(msg)
+            instance._log.warning(
+                'ALLOWING due to permitted_fail_pct={}'.format(
+                    instance.permitted_fail_pct))
+
 
 # Prevent nosetests from running experiment_test directly as a test case
 experiment_test.__test__ = False