diff --git a/coriolis/tests/integration/base.py b/coriolis/tests/integration/base.py
index 58407efe..5efec403 100644
--- a/coriolis/tests/integration/base.py
+++ b/coriolis/tests/integration/base.py
@@ -390,7 +390,7 @@ def setUp(self):
         # Safety-net cleanup for destination devices allocated by the provider.
         # Must be registered after the transfer, so it runs (LIFO) before the
         # transfer delete, while the volumes_info is still in the DB.
-        self.addCleanup(self._cleanup_provider_dst_devices)
+        self.addCleanup(self._cleanup_provider_dst_devices, self._transfer.id)
 
         # mock a few commands that are going to be ran through ssh; they won't
         # pass anyway.
@@ -421,25 +421,25 @@ def _dst_device(self):
 
         return None
 
-    def _cleanup_provider_dst_devices(self):
-        """Remove any devices the provider allocated for this test."""
+    def _cleanup_provider_dst_devices(self, transfer_id):
+        """Remove any devices the provider allocated for this transfer."""
         ctxt = self._get_db_context()
 
         try:
             executions = db_api.get_transfer_tasks_executions(
                 ctxt,
-                self._transfer.id,
+                transfer_id,
                 sort_keys=["number"],
                 sort_dirs=["desc"],
                 limit=1,
             )
             if executions:
-                self._cleanup_execution(self._transfer.id, executions[0].id)
+                self._cleanup_execution(transfer_id, executions[0].id)
         except Exception as ex:
             LOG.warn("Could not cancel execution during cleanup. Ex: %s", ex)
 
         try:
-            deletion = self._client.transfers.delete_disks(self._transfer.id)
+            deletion = self._client.transfers.delete_disks(transfer_id)
             self.wait_for_execution(deletion.id, timeout=60)
         except Exception as ex:
             LOG.warn("Could not clean up provider dest devices. Ex: %s", ex)
@@ -451,15 +451,18 @@ def _execute_and_wait(self, transfer_id, timeout=600):
         )
         self.assertExecutionCompleted(execution.id, timeout=timeout)
 
-    def _execute_concurrently_and_wait(self, transfer_ids, timeout=600):
-        """Start one execution per transfer id before waiting on any."""
-        executions = [
+    def _start_executions(self, transfer_ids):
+        """Start one execution per transfer id without waiting on any."""
+        return [
             self._client.transfer_executions.create(
                 transfer_id, shutdown_instances=False
             )
             for transfer_id in transfer_ids
         ]
-        for execution in executions:
+
+    def _execute_concurrently_and_wait(self, transfer_ids, timeout=600):
+        """Start one execution per transfer id before waiting on any."""
+        for execution in self._start_executions(transfer_ids):
             self.assertExecutionCompleted(execution.id, timeout=timeout)
 
     def _execute_transfer_and_deployment(self, deployment_kwargs=None):
@@ -745,6 +748,14 @@ def assertExecutionCompleted(self, execution_id, timeout=600):
         self.assertPoolAllocated(self._dst_pool_id)
         self.assertMachinesAvailable(self._dst_pool_id)
 
+    def _execute_concurrently_and_wait(self, transfer_ids, timeout=600):
+        # Machines allocated for a still-running execution are not AVAILABLE yet,
+        # so only check the pool once every execution is done.
+        for execution in self._start_executions(transfer_ids):
+            super().assertExecutionCompleted(execution.id, timeout=timeout)
+        self.assertPoolAllocated(self._dst_pool_id)
+        self.assertMachinesAvailable(self._dst_pool_id)
+
     def assertDeploymentCompleted(self, deployment_id, timeout=600):
         super().assertDeploymentCompleted(deployment_id, timeout=timeout)
         self.assertPoolAllocated(self._dst_pool_id)
@@ -771,3 +782,11 @@ def assertExecutionCompleted(self, execution_id, timeout=600):
         super().assertExecutionCompleted(execution_id, timeout=timeout)
         self.assertPoolAllocated(self._src_pool_id)
         self.assertMachinesAvailable(self._src_pool_id)
+
+    def _execute_concurrently_and_wait(self, transfer_ids, timeout=600):
+        # Machines allocated for a still-running execution are not AVAILABLE yet,
+        # so only check the pool once every execution is done.
+        for execution in self._start_executions(transfer_ids):
+            super().assertExecutionCompleted(execution.id, timeout=timeout)
+        self.assertPoolAllocated(self._src_pool_id)
+        self.assertMachinesAvailable(self._src_pool_id)
diff --git a/coriolis/tests/integration/test_minion_pools.py b/coriolis/tests/integration/test_minion_pools.py
index 13c003fc..cd91d987 100644
--- a/coriolis/tests/integration/test_minion_pools.py
+++ b/coriolis/tests/integration/test_minion_pools.py
@@ -236,6 +236,8 @@ def setUp(self):
             destination_minion_pool_id=self._dst_pool_id,
             origin_minion_pool_id=self._src_pool_id,
         )
+        # Registered after the transfer, so it runs (LIFO) before the transfer delete.
+        self.addCleanup(self._cleanup_provider_dst_devices, self._pool_transfer_b.id)
 
     def _wait_for_power_status(self, status, timeout=120):
         """Poll until one of the pool's machines reaches *status*."""
@@ -357,6 +359,8 @@ def setUp(self):
             destination_minion_pool_id=self._dst_pool_id,
             origin_minion_pool_id=self._src_pool_id,
         )
+        # Registered after the transfer, so it runs (LIFO) before the transfer delete.
+        self.addCleanup(self._cleanup_provider_dst_devices, self._pool_transfer_b.id)
 
     def test_excess_pool_machine_deleted_on_refresh(self):
         transfer_ids = [self._transfer.id, self._pool_transfer_b.id]