From 626079a3e3227a0b00934ce5c30d8110751dca16 Mon Sep 17 00:00:00 2001 From: Vitalii Koriakov Date: Mon, 22 Jan 2018 15:21:57 +0200 Subject: Test MetaData Self-Heal (heal command) Change-Id: I32fefdab769e5a361e4dcb5f1328b2c8da2e4f1a Signed-off-by: Vitalii Koriakov --- tests/functional/afr/heal/test_self_heal.py | 576 ++++++++++++++++++++++------ 1 file changed, 461 insertions(+), 115 deletions(-) (limited to 'tests/functional/afr') diff --git a/tests/functional/afr/heal/test_self_heal.py b/tests/functional/afr/heal/test_self_heal.py index c378af8ee..d7506a654 100755 --- a/tests/functional/afr/heal/test_self_heal.py +++ b/tests/functional/afr/heal/test_self_heal.py @@ -116,13 +116,6 @@ class TestSelfHeal(GlusterBaseClass): raise ExecutionError("IO failed on some of the clients") g.log.info("IO is successful on all mounts") - # List all files and dirs created - g.log.info("List all files and directories:") - ret = list_all_files_and_dirs_mounts(self.mounts) - if not ret: - raise ExecutionError("Failed to list all files and dirs") - g.log.info("Listing all files and directories is successful") - # Cleanup and umount volume g.log.info("Starting to Unmount Volume and Cleanup Volume") ret = self.unmount_volume_and_cleanup_volume(mounts=self.mounts) @@ -143,12 +136,12 @@ class TestSelfHeal(GlusterBaseClass): "entry-self-heal": "off" "data-self-heal": "off" - create IO - - Get areequal before getting bricks offline + - Get arequal before getting bricks offline - set the volume option "self-heal-daemon": "off" - bring down all bricks processes from selected set - - Get areequal after getting bricks offline and compare with - areequal before getting bricks offline + - Get areeual after getting bricks offline and compare with + arequal before getting bricks offline - modify the data - bring bricks online - set the volume option @@ -196,11 +189,11 @@ class TestSelfHeal(GlusterBaseClass): self.io_validation_complete = True g.log.info("IO is successful on all mounts") - # Get areequal before getting bricks offline - g.log.info('Getting areequal before getting bricks offline...') + # Get arequal before getting bricks offline + g.log.info('Getting arequal before getting bricks offline...') ret, result_before_offline = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal before getting bricks offline ' + g.log.info('Getting arequal before getting bricks offline ' 'is successful') # Setting options @@ -231,14 +224,14 @@ class TestSelfHeal(GlusterBaseClass): g.log.info('Bringing bricks %s offline is successful', bricks_to_bring_offline) - # Get areequal after getting bricks offline - g.log.info('Getting areequal after getting bricks offline...') + # Get arequal after getting bricks offline + g.log.info('Getting arequal after getting bricks offline...') ret, result_after_offline = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal after getting bricks offline ' + g.log.info('Getting arequal after getting bricks offline ' 'is successful') - # Checking areequals before bringing bricks offline + # Checking arequals before bringing bricks offline # and after bringing bricks offline self.assertEqual(result_before_offline, result_after_offline, 'Checksums before and ' @@ -422,21 +415,21 @@ class TestSelfHeal(GlusterBaseClass): "entry-self-heal": "off" "data-self-heal": "off" - create IO - - get areequal before getting bricks offline + - get arequal before getting bricks offline - set the volume option "self-heal-daemon": "off" - bring down all bricks processes from selected set - - get areequal after getting bricks offline and compare with + - get arequal after getting bricks offline and compare with arequal after bringing bricks offline - modify the data - - get areequal before getting bricks online + - get arequal before getting bricks online - bring bricks online - set the volume option "self-heal-daemon": "on" - check daemons and start healing - check if heal is completed - check for split-brain - - get areequal after getting bricks online and compare with + - get arequal after getting bricks online and compare with arequal before bringing bricks online """ # pylint: disable=too-many-statements @@ -491,11 +484,11 @@ class TestSelfHeal(GlusterBaseClass): "python %s delete %s"] for cmd in cmd_list: - # Get areequal before getting bricks offline - g.log.info('Getting areequal before getting bricks offline...') + # Get arequal before getting bricks offline + g.log.info('Getting arequal before getting bricks offline...') ret, result_before_offline = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal before getting bricks offline ' + g.log.info('Getting arequal before getting bricks offline ' 'is successful') # Setting options @@ -528,14 +521,14 @@ class TestSelfHeal(GlusterBaseClass): g.log.info('Bringing bricks %s offline is successful', bricks_to_bring_offline) - # Get areequal after getting bricks offline - g.log.info('Getting areequal after getting bricks offline...') + # Get arequal after getting bricks offline + g.log.info('Getting arequal after getting bricks offline...') ret, result_after_offline = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal after getting bricks offline ' + g.log.info('Getting arequal after getting bricks offline ' 'is successful') - # Checking areequals before bringing bricks offline + # Checking arequals before bringing bricks offline # and after bringing bricks offline self.assertEqual(result_before_offline, result_after_offline, 'Checksums are not equal') @@ -563,11 +556,11 @@ class TestSelfHeal(GlusterBaseClass): self.io_validation_complete = True g.log.info("IO is successful on all mounts") - # Get areequal before getting bricks online - g.log.info('Getting areequal before getting bricks online...') + # Get arequal before getting bricks online + g.log.info('Getting arequal before getting bricks online...') ret, result_before_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal before getting bricks online ' + g.log.info('Getting arequal before getting bricks online ' 'is successful') # List all files and dirs created @@ -636,11 +629,11 @@ class TestSelfHeal(GlusterBaseClass): self.assertFalse(ret, 'Volume is in split-brain state') g.log.info('Volume is not in split-brain state') - # Get areequal after getting bricks online - g.log.info('Getting areequal after getting bricks online...') + # Get arequal after getting bricks online + g.log.info('Getting arequal after getting bricks online...') ret, result_after_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal after getting bricks online ' + g.log.info('Getting arequal after getting bricks online ' 'is successful') # List all files and dirs created @@ -650,7 +643,7 @@ class TestSelfHeal(GlusterBaseClass): raise ExecutionError("Failed to list all files and dirs") g.log.info("Listing all files and directories is successful") - # Checking areequals before bringing bricks online + # Checking arequals before bringing bricks online # and after bringing bricks online self.assertEqual(result_before_online, result_after_online, 'Checksums are not equal') @@ -669,10 +662,10 @@ class TestSelfHeal(GlusterBaseClass): "data-self-heal-algorithm": "diff" "self-heal-daemon": "off" - create IO - - calculate areequal + - calculate arequal - bring down all bricks processes from selected set - modify the data - - get areequal before getting bricks online + - get arequal before getting bricks online - bring bricks online - expand volume by adding bricks to the volume - do rebalance @@ -680,7 +673,7 @@ class TestSelfHeal(GlusterBaseClass): - start healing - check if heal is completed - check for split-brain - - calculate areequal and compare with arequal before bringing bricks + - calculate arequal and compare with arequal before bringing bricks offline and after bringing bricks online """ # pylint: disable=too-many-branches,too-many-statements @@ -767,11 +760,11 @@ class TestSelfHeal(GlusterBaseClass): self.assertTrue(ret, "IO failed on some of the clients") g.log.info("IO is successful on all mounts") - # Get areequal before getting bricks online - g.log.info('Getting areequal before getting bricks online...') + # Get arequal before getting bricks online + g.log.info('Getting arequal before getting bricks online...') ret, result_before_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal before getting bricks online ' + g.log.info('Getting arequal before getting bricks online ' 'is successful') # Bring brick online @@ -848,14 +841,14 @@ class TestSelfHeal(GlusterBaseClass): self.assertFalse(ret, 'Volume is in split-brain state') g.log.info('Volume is not in split-brain state') - # Get areequal after getting bricks online - g.log.info('Getting areequal after getting bricks online...') + # Get arequal after getting bricks online + g.log.info('Getting arequal after getting bricks online...') ret, result_after_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal after getting bricks online ' + g.log.info('Getting arequal after getting bricks online ' 'is successful') - # Checking areequals before bringing bricks offline + # Checking arequals before bringing bricks offline # and after bringing bricks online self.assertItemsEqual(result_before_online, result_after_online, 'Checksums are not equal') @@ -870,10 +863,10 @@ class TestSelfHeal(GlusterBaseClass): - create IO - bring down all bricks processes from selected set - modify the data - - calculate areequal + - calculate arequal - bring bricks online - start healing - - calculate areequal and compare with arequal before bringing bricks + - calculate arequal and compare with arequal before bringing bricks offline and after bringing bricks online """ # pylint: disable=too-many-locals,too-many-statements @@ -944,11 +937,11 @@ class TestSelfHeal(GlusterBaseClass): self.assertTrue(ret, "IO failed on some of the clients") g.log.info("IO is successful on all mounts") - # Get areequal before getting bricks online - g.log.info('Getting areequal before getting bricks online...') + # Get arequal before getting bricks online + g.log.info('Getting arequal before getting bricks online...') ret, result_before_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal before getting bricks online ' + g.log.info('Getting arequal before getting bricks online ' 'is successful') # Bring brick online @@ -995,14 +988,14 @@ class TestSelfHeal(GlusterBaseClass): self.assertFalse(ret, 'Volume is in split-brain state') g.log.info('Volume is not in split-brain state') - # Get areequal after getting bricks online - g.log.info('Getting areequal after getting bricks online...') + # Get arequal after getting bricks online + g.log.info('Getting arequal after getting bricks online...') ret, result_after_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal after getting bricks online ' + g.log.info('Getting arequal after getting bricks online ' 'is successful') - # Checking areequals before bringing bricks online + # Checking arequals before bringing bricks online # and after bringing bricks online self.assertItemsEqual(result_before_online, result_after_online, 'Checksums are not equal') @@ -1018,10 +1011,10 @@ class TestSelfHeal(GlusterBaseClass): - create IO - bring down all bricks processes from selected set - modify the data - - calculate areequal + - calculate arequal - bring bricks online - start healing - - calculate areequal and compare with arequal before bringing bricks + - calculate arequal and compare with arequal before bringing bricks offline and after bringing bricks online """ # pylint: disable=too-many-locals,too-many-statements @@ -1092,11 +1085,11 @@ class TestSelfHeal(GlusterBaseClass): self.assertTrue(ret, "IO failed on some of the clients") g.log.info("IO is successful on all mounts") - # Get areequal before getting bricks online - g.log.info('Getting areequal before getting bricks online...') + # Get arequal before getting bricks online + g.log.info('Getting arequal before getting bricks online...') ret, result_before_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal before getting bricks online ' + g.log.info('Getting arequal before getting bricks online ' 'is successful') # Bring brick online @@ -1143,14 +1136,14 @@ class TestSelfHeal(GlusterBaseClass): self.assertFalse(ret, 'Volume is in split-brain state') g.log.info('Volume is not in split-brain state') - # Get areequal after getting bricks online - g.log.info('Getting areequal after getting bricks online...') + # Get arequal after getting bricks online + g.log.info('Getting arequal after getting bricks online...') ret, result_after_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal after getting bricks online ' + g.log.info('Getting arequal after getting bricks online ' 'is successful') - # Checking areequals before bringing bricks online + # Checking arequals before bringing bricks online # and after bringing bricks online self.assertItemsEqual(result_before_online, result_after_online, 'Checksums are not equal') @@ -1164,16 +1157,16 @@ class TestSelfHeal(GlusterBaseClass): Description: - create IO - - calculate areequal + - calculate arequal - bring down all bricks processes from selected set - calculate arequal and compare with arequal before getting bricks offline - modify the data - - areequal before getting bricks online + - arequal before getting bricks online - bring bricks online - check daemons and healing completion - start healing - - calculate areequal and compare with arequal before bringing bricks + - calculate arequal and compare with arequal before bringing bricks online and after bringing bricks online """ # pylint: disable=too-many-locals,too-many-statements @@ -1213,11 +1206,11 @@ class TestSelfHeal(GlusterBaseClass): self.assertTrue(ret, "IO failed on some of the clients") g.log.info("IO is successful on all mounts") - # Get areequal before getting bricks offline - g.log.info('Getting areequal before getting bricks offline...') + # Get arequal before getting bricks offline + g.log.info('Getting arequal before getting bricks offline...') ret, result_before_offline = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal before getting bricks offline ' + g.log.info('Getting arequal before getting bricks offline ' 'is successful') # Select bricks to bring offline @@ -1241,14 +1234,14 @@ class TestSelfHeal(GlusterBaseClass): g.log.info('Bringing bricks %s offline is successful', bricks_to_bring_offline) - # Get areequal after getting bricks offline - g.log.info('Getting areequal after getting bricks offline...') + # Get arequal after getting bricks offline + g.log.info('Getting arequal after getting bricks offline...') ret, result_after_offline = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal after getting bricks offline ' + g.log.info('Getting arequal after getting bricks offline ' 'is successful') - # Checking areequals before bringing bricks offline + # Checking arequals before bringing bricks offline # and after bringing bricks offline self.assertItemsEqual(result_before_offline, result_after_offline, 'Checksums before and after ' @@ -1284,11 +1277,11 @@ class TestSelfHeal(GlusterBaseClass): self.assertTrue(ret, "IO failed on some of the clients") g.log.info("IO is successful on all mounts") - # Get areequal before getting bricks online - g.log.info('Getting areequal before getting bricks online...') + # Get arequal before getting bricks online + g.log.info('Getting arequal before getting bricks online...') ret, result_before_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal before getting bricks online ' + g.log.info('Getting arequal before getting bricks online ' 'is successful') # Bring brick online @@ -1335,14 +1328,14 @@ class TestSelfHeal(GlusterBaseClass): self.assertFalse(ret, 'Volume is in split-brain state') g.log.info('Volume is not in split-brain state') - # Get areequal after getting bricks online - g.log.info('Getting areequal after getting bricks online...') + # Get arequal after getting bricks online + g.log.info('Getting arequal after getting bricks online...') ret, result_after_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal after getting bricks online ' + g.log.info('Getting arequal after getting bricks online ' 'is successful') - # Checking areequals before bringing bricks online + # Checking arequals before bringing bricks online # and after bringing bricks online self.assertItemsEqual(result_before_online, result_after_online, 'Checksums before and ' @@ -1362,20 +1355,20 @@ class TestSelfHeal(GlusterBaseClass): "data-self-heal-algorithm": "diff" "self-heal-daemon": "off" - create IO - - calculate areequal + - calculate arequal - bring down all bricks processes from selected set - - calculate areequals and compare with arequal + - calculate arequals and compare with arequal before bringing bricks offline - modify the data and verify whether the links are properly created - - calculate areequal before getting bricks online + - calculate arequal before getting bricks online - bring bricks online - set the volume option "self-heal-daemon": "on" - check daemons and start healing - check is heal is complited - check for split-brain - - calculate areequal after getting bricks online and compare with - areequal before getting bricks online + - calculate arequal after getting bricks online and compare with + arequal before getting bricks online """ # pylint: disable=too-many-locals,too-many-statements # Setting options @@ -1424,11 +1417,11 @@ class TestSelfHeal(GlusterBaseClass): self.assertTrue(ret, "IO failed on some of the clients") g.log.info("IO is successful on all mounts") - # Get areequal before getting bricks offline - g.log.info('Getting areequal before getting bricks offline...') + # Get arequal before getting bricks offline + g.log.info('Getting arequal before getting bricks offline...') ret, result_before_offline = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal before getting bricks offline ' + g.log.info('Getting arequal before getting bricks offline ' 'is successful') # Select bricks to bring offline @@ -1452,14 +1445,14 @@ class TestSelfHeal(GlusterBaseClass): g.log.info('Bringing bricks %s offline is successful', bricks_to_bring_offline) - # Get areequal after getting bricks offline - g.log.info('Getting areequal after getting bricks offline...') + # Get arequal after getting bricks offline + g.log.info('Getting arequal after getting bricks offline...') ret, result_after_offline = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal after getting bricks offline ' + g.log.info('Getting arequal after getting bricks offline ' 'is successful') - # Checking areequals before bringing bricks offline + # Checking arequals before bringing bricks offline # and after bringing bricks offline self.assertItemsEqual(result_before_offline, result_after_offline, 'Checksums before and ' @@ -1517,11 +1510,11 @@ class TestSelfHeal(GlusterBaseClass): g.log.info('Links for %s are properly created', mount_object.mountpoint) - # Get areequal before getting bricks online - g.log.info('Getting areequal before getting bricks online...') + # Get arequal before getting bricks online + g.log.info('Getting arequal before getting bricks online...') ret, result_before_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal before getting bricks online ' + g.log.info('Getting arequal before getting bricks online ' 'is successful') # Bring brick online @@ -1580,14 +1573,14 @@ class TestSelfHeal(GlusterBaseClass): self.assertFalse(ret, 'Volume is in split-brain state') g.log.info('Volume is not in split-brain state') - # Get areequal after getting bricks online - g.log.info('Getting areequal after getting bricks online...') + # Get arequal after getting bricks online + g.log.info('Getting arequal after getting bricks online...') ret, result_after_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal after getting bricks online ' + g.log.info('Getting arequal after getting bricks online ' 'is successful') - # Checking areequals before bringing bricks online + # Checking arequals before bringing bricks online # and after bringing bricks online self.assertItemsEqual(result_before_online, result_after_online, 'Checksums before and ' @@ -1606,7 +1599,7 @@ class TestSelfHeal(GlusterBaseClass): "self-heal-daemon": "off" - bring down all bricks processes from selected set - create IO (50k files) - - Get areequal before getting bricks online + - Get arequal before getting bricks online - bring bricks online - set the volume option "self-heal-daemon": "on" @@ -1614,12 +1607,12 @@ class TestSelfHeal(GlusterBaseClass): - start healing - check if heal is completed - check for split-brain - - get areequal after getting bricks online and compare with - areequal before getting bricks online + - get arequal after getting bricks online and compare with + arequal before getting bricks online - add bricks - do rebalance - - get areequal after adding bricks and compare with - areequal after getting bricks online + - get arequal after adding bricks and compare with + arequal after getting bricks online """ # pylint: disable=too-many-locals,too-many-statements # Setting options @@ -1673,11 +1666,11 @@ class TestSelfHeal(GlusterBaseClass): self.assertTrue(ret, "IO failed on some of the clients") g.log.info("IO is successful on all mounts") - # Get areequal before getting bricks online - g.log.info('Getting areequal before getting bricks online...') + # Get arequal before getting bricks online + g.log.info('Getting arequal before getting bricks online...') ret, result_before_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal before getting bricks online ' + g.log.info('Getting arequal before getting bricks online ' 'is successful') # Bring brick online @@ -1737,14 +1730,14 @@ class TestSelfHeal(GlusterBaseClass): self.assertFalse(ret, 'Volume is in split-brain state') g.log.info('Volume is not in split-brain state') - # Get areequal after getting bricks online - g.log.info('Getting areequal after getting bricks online...') + # Get arequal after getting bricks online + g.log.info('Getting arequal after getting bricks online...') ret, result_after_online = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal after getting bricks online ' + g.log.info('Getting arequal after getting bricks online ' 'is successful') - # Checking areequals before bringing bricks online + # Checking arequals before bringing bricks online # and after bringing bricks online self.assertItemsEqual(result_before_online, result_after_online, 'Checksums before and ' @@ -1769,14 +1762,14 @@ class TestSelfHeal(GlusterBaseClass): self.assertTrue(ret, 'Rebalance is not completed') g.log.info('Rebalance is completed successfully') - # Get areequal after adding bricks - g.log.info('Getting areequal after adding bricks...') + # Get arequal after adding bricks + g.log.info('Getting arequal after adding bricks...') ret, result_after_adding_bricks = collect_mounts_arequal(self.mounts) self.assertTrue(ret, 'Failed to get arequal') - g.log.info('Getting areequal after getting bricks ' + g.log.info('Getting arequal after getting bricks ' 'is successful') - # Checking areequals after bringing bricks online + # Checking arequals after bringing bricks online # and after adding bricks self.assertItemsEqual(result_after_online, result_after_adding_bricks, 'Checksums after bringing bricks online and ' @@ -1891,3 +1884,356 @@ class TestSelfHeal(GlusterBaseClass): # Comparing the results g.log.info("comparing both the results") self.assertEqual(result_before, result_after, "Arequals are not equal") + + +@runs_on([['replicated', 'distributed-replicated'], + ['glusterfs', 'cifs', 'nfs']]) +class TestMetadataSelfHeal(GlusterBaseClass): + """ + Description: + Test cases related to metadata delf heal + in default configuration of the volume + """ + + @classmethod + def setUpClass(cls): + # Calling GlusterBaseClass setUpClass + GlusterBaseClass.setUpClass.im_func(cls) + + # Upload io scripts for running IO on mounts + g.log.info("Upload io scripts to clients %s for running IO on mounts", + cls.clients) + script_local_path = ("/usr/share/glustolibs/io/scripts/" + "file_dir_ops.py") + cls.script_upload_path = ("/usr/share/glustolibs/io/scripts/" + "file_dir_ops.py") + ret = upload_scripts(cls.clients, [script_local_path]) + if not ret: + raise ExecutionError("Failed to upload IO scripts to clients %s" + % cls.clients) + g.log.info("Successfully uploaded IO scripts to clients %s", + cls.clients) + + cls.counter = 1 + # int: Value of counter is used for dirname-start-num argument for + # file_dir_ops.py create_deep_dirs_with_files. + + # The --dir-length argument value for file_dir_ops.py + # create_deep_dirs_with_files is set to 10 (refer to the cmd in setUp + # method). This means every mount will create + # 10 top level dirs. For every mountpoint/testcase to create new set of + # dirs, we are incrementing the counter by --dir-length value i.e 10 + # in this test suite. + + # If we are changing the --dir-length to new value, ensure the counter + # is also incremented by same value to create new set of files/dirs. + + def setUp(self): + # Calling GlusterBaseClass setUp + GlusterBaseClass.setUp.im_func(self) + + self.all_mounts_procs = [] + self.io_validation_complete = False + + for mount_object in self.mounts: + # Create user qa + g.log.info("Creating user 'qa'...") + command = "useradd qa" + ret, _, err = g.run(mount_object.client_system, command) + + if 'already exists' in err: + g.log.warn("User 'qa' is already exists") + else: + g.log.info("User 'qa' is created successfully") + + # Setup Volume and Mount Volume + g.log.info("Starting to Setup Volume and Mount Volume") + ret = self.setup_volume_and_mount_volume(mounts=self.mounts) + if not ret: + raise ExecutionError("Failed to Setup_Volume and Mount_Volume") + g.log.info("Successful in Setup Volume and Mount Volume") + + def tearDown(self): + """ + If test method failed before validating IO, tearDown waits for the + IO's to complete and checks for the IO exit status + + Cleanup and umount volume + """ + if not self.io_validation_complete: + g.log.info("Wait for IO to complete as IO validation did not " + "succeed in test method") + ret = wait_for_io_to_complete(self.all_mounts_procs, self.mounts) + if not ret: + raise ExecutionError("IO failed on some of the clients") + g.log.info("IO is successful on all mounts") + + for mount_object in self.mounts: + # Delete user + g.log.info('Deleting user qa...') + command = "userdel -r qa" + ret, _, err = g.run(mount_object.client_system, command) + + if 'does not exist' in err: + g.log.warn('User qa is already deleted') + else: + g.log.info('User qa successfully deleted') + + # Cleanup and umount volume + g.log.info("Starting to Unmount Volume and Cleanup Volume") + ret = self.unmount_volume_and_cleanup_volume(mounts=self.mounts) + if not ret: + raise ExecutionError("Failed to umount the vol & cleanup Volume") + g.log.info("Successful in umounting the volume and Cleanup") + + # Calling GlusterBaseClass teardown + GlusterBaseClass.tearDown.im_func(self) + + def test_metadata_self_heal(self): + """ + Test MetaData Self-Heal (heal command) + + Description: + - set the volume option + "metadata-self-heal": "off" + "entry-self-heal": "off" + "data-self-heal": "off" + - create IO + - set the volume option + "self-heal-daemon": "off" + - bring down all bricks processes from selected set + - Change the permissions, ownership and the group + of the files under "test_meta_data_self_heal" folder + - get arequal before getting bricks online + - bring bricks online + - set the volume option + "self-heal-daemon": "on" + - check daemons and start healing + - check is heal is completed + - check for split-brain + - get arequal after getting bricks online and compare with + arequal before getting bricks online + - check group and user are 'qa' + """ + # pylint: disable=too-many-locals,too-many-statements + # Setting options + g.log.info('Setting options...') + options = {"metadata-self-heal": "off", + "entry-self-heal": "off", + "data-self-heal": "off"} + ret = set_volume_options(self.mnode, self.volname, options) + self.assertTrue(ret, 'Failed to set options') + g.log.info("Options " + "'metadata-self-heal', " + "'entry-self-heal', " + "'data-self-heal', " + "are set to 'off' successfully") + + # Creating files on client side + test_meta_data_self_heal_folder = 'test_meta_data_self_heal' + for mount_object in self.mounts: + g.log.info("Generating data for %s:%s", + mount_object.client_system, mount_object.mountpoint) + + # Create files + g.log.info('Creating files...') + command = ("cd %s/ ; " + "mkdir %s ;" + "cd %s/ ;" + "for i in `seq 1 50` ; " + "do dd if=/dev/urandom of=test.$i bs=10k count=1 ; " + "done ;" + % (mount_object.mountpoint, + test_meta_data_self_heal_folder, + test_meta_data_self_heal_folder)) + + proc = g.run_async(mount_object.client_system, command, + user=mount_object.user) + self.all_mounts_procs.append(proc) + + # Validate IO + g.log.info("Wait for IO to complete and validate IO ...") + ret = validate_io_procs(self.all_mounts_procs, self.mounts) + self.io_validation_complete = True + self.assertTrue(ret, "IO failed on some of the clients") + g.log.info("IO is successful on all mounts") + + # Setting options + g.log.info('Setting options...') + options = {"self-heal-daemon": "off"} + ret = set_volume_options(self.mnode, self.volname, options) + self.assertTrue(ret, 'Failed to set options') + g.log.info("Option 'self-heal-daemon' is set to 'off' successfully") + + # Select bricks to bring offline + bricks_to_bring_offline_dict = (select_bricks_to_bring_offline( + self.mnode, self.volname)) + bricks_to_bring_offline = filter(None, ( + bricks_to_bring_offline_dict['hot_tier_bricks'] + + bricks_to_bring_offline_dict['cold_tier_bricks'] + + bricks_to_bring_offline_dict['volume_bricks'])) + + # Bring brick offline + g.log.info('Bringing bricks %s offline...', bricks_to_bring_offline) + ret = bring_bricks_offline(self.volname, bricks_to_bring_offline) + self.assertTrue(ret, 'Failed to bring bricks %s offline' % + bricks_to_bring_offline) + + ret = are_bricks_offline(self.mnode, self.volname, + bricks_to_bring_offline) + self.assertTrue(ret, 'Bricks %s are not offline' + % bricks_to_bring_offline) + g.log.info('Bringing bricks %s offline is successful', + bricks_to_bring_offline) + + # Changing the permissions, ownership and the group + # of the files under "test_meta_data_self_heal" folder + for mount_object in self.mounts: + g.log.info("Modifying data for %s:%s", + mount_object.client_system, mount_object.mountpoint) + + # Change permissions to 444 + g.log.info('Changing permissions...') + command = ("cd %s/%s/ ; " + "chmod -R 444 *" + % (mount_object.mountpoint, + test_meta_data_self_heal_folder)) + ret, out, err = g.run(mount_object.client_system, command) + self.assertEqual(ret, 0, err) + g.log.info('Permissions are changed successfully') + + # Change the ownership to qa + g.log.info('Changing the ownership...') + command = ("cd %s/%s/ ; " + "chown -R qa *" + % (mount_object.mountpoint, + test_meta_data_self_heal_folder)) + ret, out, err = g.run(mount_object.client_system, command) + self.assertEqual(ret, 0, err) + g.log.info('Ownership is changed successfully') + + # Change the group to qa + g.log.info('Changing the group...') + command = ("cd %s/%s/ ; " + "chgrp -R qa *" + % (mount_object.mountpoint, + test_meta_data_self_heal_folder)) + ret, out, err = g.run(mount_object.client_system, command) + self.assertEqual(ret, 0, err) + g.log.info('Group is changed successfully') + + # Get arequal before getting bricks online + g.log.info('Getting arequal before getting bricks online...') + ret, result_before_online = collect_mounts_arequal(self.mounts) + self.assertTrue(ret, 'Failed to get arequal') + g.log.info('Getting arequal before getting bricks online ' + 'is successful') + + # Bring brick online + g.log.info('Bringing bricks %s online...', bricks_to_bring_offline) + ret = bring_bricks_online(self.mnode, self.volname, + bricks_to_bring_offline) + self.assertTrue(ret, 'Failed to bring bricks %s online' % + bricks_to_bring_offline) + g.log.info('Bringing bricks %s online is successful', + bricks_to_bring_offline) + + # Setting options + g.log.info('Setting options...') + options = {"self-heal-daemon": "on"} + ret = set_volume_options(self.mnode, self.volname, options) + self.assertTrue(ret, 'Failed to set options') + g.log.info("Option 'self-heal-daemon' is set to 'on' successfully") + + # Wait for volume processes to be online + g.log.info("Wait for volume processes to be online") + ret = wait_for_volume_process_to_be_online(self.mnode, self.volname) + self.assertTrue(ret, ("Volume process %s not online " + "despite waiting for 5 minutes", self.volname)) + g.log.info("Successful in waiting for volume %s processes to be " + "online", self.volname) + + # Verify volume's all process are online + g.log.info("Verifying volume's all process are online") + ret = verify_all_process_of_volume_are_online(self.mnode, self.volname) + self.assertTrue(ret, ("Volume %s : All process are not online" + % self.volname)) + g.log.info("Volume %s : All process are online", self.volname) + + # Wait for self-heal-daemons to be online + g.log.info("Waiting for self-heal-daemons to be online") + ret = is_shd_daemonized(self.all_servers) + self.assertTrue(ret, "Either No self heal daemon process found") + g.log.info("All self-heal-daemons are online") + + # Start healing + ret = trigger_heal(self.mnode, self.volname) + self.assertTrue(ret, 'Heal is not started') + g.log.info('Healing is started') + + # Monitor heal completion + ret = monitor_heal_completion(self.mnode, self.volname) + self.assertTrue(ret, 'Heal has not yet completed') + + # Check if heal is completed + ret = is_heal_complete(self.mnode, self.volname) + self.assertTrue(ret, 'Heal is not complete') + g.log.info('Heal is completed successfully') + + # Check for split-brain + ret = is_volume_in_split_brain(self.mnode, self.volname) + self.assertFalse(ret, 'Volume is in split-brain state') + g.log.info('Volume is not in split-brain state') + + # Get arequal after getting bricks online + g.log.info('Getting arequal after getting bricks online...') + ret, result_after_online = collect_mounts_arequal(self.mounts) + self.assertTrue(ret, 'Failed to get arequal') + g.log.info('Getting arequal after getting bricks online ' + 'is successful') + + # Checking arequals before bringing bricks online + # and after bringing bricks online + self.assertItemsEqual(result_before_online, result_after_online, + 'Checksums are not equal') + g.log.info('Checksums before bringing bricks online ' + 'and after bringing bricks online are equal') + + # Check for user and group + for mount_object in self.mounts: + # Get file list + command = ("cd %s/%s/ ; " + "ls" + % (mount_object.mountpoint, + test_meta_data_self_heal_folder)) + ret, out, err = g.run(mount_object.client_system, command) + file_list = out.split() + + # Checking for user and group + g.log.info('Checking for user and group...') + conn = g.rpyc_get_connection(mount_object.client_system) + if conn is None: + raise Exception("Unable to get connection on node %s" % + mount_object.client_system) + + for file_name in file_list: + file_to_check = '%s/%s/%s' % (mount_object.mountpoint, + test_meta_data_self_heal_folder, + file_name) + + g.log.info('Checking for user and group for %s...', file_name) + # Check for user + uid = conn.modules.os.stat(file_to_check).st_uid + username = conn.modules.pwd.getpwuid(uid).pw_name + self.assertEqual(username, 'qa', 'User %s is not equal qa' + % username) + g.log.info("User is 'qa' for %s", file_name) + + # Check for group + gid = conn.modules.os.stat(file_to_check).st_gid + groupname = conn.modules.grp.getgrgid(gid).gr_name + self.assertEqual(groupname, 'qa', 'Group %s is not equal qa' + % groupname) + g.log.info("Group is 'qa' for %s", file_name) + + g.rpyc_close_connection(host=mount_object.client_system) -- cgit