Merge pull request #8042 from VJalili/os_tests

Extend ObjectStore integration tests

test/integration/objectstore/test_jobs.py

@@ -34,6 +34,8 @@ DISTRIBUTED_OBJECT_STORE_CONFIG_TEMPLATE = string.Template("""<?xml version="1.0
 </object_store>
 """)
 
+TEST_INPUT_FILES_CONTENT = "1 2 3"
+
 
 class ObjectStoreJobsIntegrationTestCase(integration_util.IntegrationTestCase):
 
@@ -55,10 +57,8 @@ class ObjectStoreJobsIntegrationTestCase(integration_util.IntegrationTestCase):
     def setUp(self):
         super(ObjectStoreJobsIntegrationTestCase, self).setUp()
         self.dataset_populator = DatasetPopulator(self.galaxy_interactor)
-
-    def test_tool_simple_constructs(self):
         with self.dataset_populator.test_history() as history_id:
-            hda1 = self.dataset_populator.new_dataset(history_id, content="1 2 3")
+            hda1 = self.dataset_populator.new_dataset(history_id, content=TEST_INPUT_FILES_CONTENT)
             create_10_inputs = {
                 "input1": {"src": "hda", "id": hda1["id"]},
                 "input2": {"src": "hda", "id": hda1["id"]},
@@ -71,6 +71,25 @@ class ObjectStoreJobsIntegrationTestCase(integration_util.IntegrationTestCase):
             )
             self.dataset_populator.wait_for_history(history_id)
 
+    def test_files_count_and_content_in_each_objectstore_backend(self):
+        """
+        According to the ObjectStore configuration given in the
+        `DISTRIBUTED_OBJECT_STORE_CONFIG_TEMPLATE` variable, datasets
+        can be stored on three backends, named:
+            -   primary/files1;
+            -   primary/files2;
+            -   secondary/files3.
+
+        Objectstore _randomly_ distributes tools outputs on
+        `primary/files1` and `primary/files2`, and will use
+        `secondary/files3` if both `primary` backends fail.
+
+        This test runs a tools that creates ten dummy datasets,
+        and asserts if ObjectStore correctly creates ten files
+        in `primary/files1` and `primary/files2`, and none in
+        `secondary/files3`, assuming it will not fail persisting
+        data in `primary` backend.
+        """
         files_1_count = _files_count(self.files1_path)
         files_2_count = _files_count(self.files2_path)
         files_3_count = _files_count(self.files3_path)
@@ -79,17 +98,39 @@ class ObjectStoreJobsIntegrationTestCase(integration_util.IntegrationTestCase):
         assert files_3_count == 0
 
         # Ensure the 10 inputs were written to one of the distributed object store's disk
-        # stores (it will have either 10 or 11 depeending on whether the input was also
+        # stores (it will have either 10 or 11 depending on whether the input was also
         # written there. The other disk store may or may not have the input file so should
         # have at most one file.
-        assert (files_1_count >= 10) or (files_2_count >= 10)
-        assert (files_1_count <= 1) or (files_2_count <= 1)
+        assert (files_1_count + files_2_count == 10) or (files_1_count + files_2_count == 11)
 
         # Other sanity checks on the test - just make sure the test was setup as intended
         # and not actually testing object store behavior.
         assert (files_1_count <= 11) and (files_2_count <= 11)
         assert (files_1_count >= 0) and (files_2_count >= 0)
 
+        # TODO: ideally the following assertion should be separated in a different test method.
+        contents = []
+        path1_files = _get_datasets_files_in_path(self.files1_path)
+        path2_files = _get_datasets_files_in_path(self.files2_path)
+        path3_files = _get_datasets_files_in_path(self.files3_path)
+        for filename in path1_files + path2_files + path3_files:
+            with open(filename) as f:
+                content = f.read().strip()
+                if content != TEST_INPUT_FILES_CONTENT:
+                    contents.append(content)
+
+        for expected_content in range(1, 10):
+            assert str(expected_content) in contents
+
 
 def _files_count(directory):
     return sum(len(files) for _, _, files in os.walk(directory))
+
+
+def _get_datasets_files_in_path(directory):
+    files = []
+    for path, _, filename in os.walk(directory):
+        for f in filename:
+            if f.endswith(".dat"):
+                files.append(os.path.join(path, f))
+    return files

test/integration/objectstore/test_selection.py

@@ -9,8 +9,8 @@ from base.populators import (
 )
 
 SCRIPT_DIRECTORY = os.path.abspath(os.path.dirname(__file__))
-JOB_CONFIG_FILE = os.path.join(SCRIPT_DIRECTORY, "objectstore_selection_job_conf.xml")
-JOB_RESOURCE_PARAMETERS_CONFIG_FILE = os.path.join(SCRIPT_DIRECTORY, "objectstore_selection_job_resource_parameters_conf.xml")
+JOB_CONFIG_FILE = os.path.join(SCRIPT_DIRECTORY, "selection_job_conf.xml")
+JOB_RESOURCE_PARAMETERS_CONFIG_FILE = os.path.join(SCRIPT_DIRECTORY, "selection_job_resource_parameters_conf.xml")
 
 DISTRIBUTED_OBJECT_STORE_CONFIG_TEMPLATE = string.Template("""<?xml version="1.0"?>
 <object_store type="distributed" id="primary" order="0">
@@ -80,7 +80,6 @@ class ObjectStoreJobsIntegrationTestCase(integration_util.IntegrationTestCase):
         assert dynamic_s3 == files_dynamic_s3_count
 
     def test_tool_simple_constructs(self):
-
         with self.dataset_populator.test_history() as history_id:
 
             def _run_tool(tool_id, inputs):