Beispiel #1
0
    def patching_environment(self):
        """Apply patches on deployed environment

        Scenario:
        1. Revert snapshot of deployed environment
        2. Run Rally benchmark tests and store results
        3. Download patched packages on master node and make local repositories
        4. Add new local repositories on slave nodes
        5. Download late artifacts and clean generated images if needed
        6. Perform actions required to apply patches
        7. Verify that fix works
        8. Run OSTF
        9. Run Rally benchmark tests and compare results

        Duration 15m
        """

        # Step #1
        if not self.env.revert_snapshot(self.snapshot_name):
            raise PatchingTestException('Environment revert from snapshot "{0}'
                                        '" failed.'.format(self.snapshot_name))
        # Check that environment exists and it's ready for patching
        cluster_id = self.fuel_web.get_last_created_cluster()
        assert_is_not_none(cluster_id, 'Environment for patching not found.')

        # Step #2
        if settings.PATCHING_RUN_RALLY:
            rally_benchmarks = {}
            benchmark_results1 = {}
            for tag in set(settings.RALLY_TAGS):
                rally_benchmarks[tag] = RallyBenchmarkTest(
                    container_repo=settings.RALLY_DOCKER_REPO,
                    environment=self.env,
                    cluster_id=cluster_id,
                    test_type=tag)
                benchmark_results1[tag] = rally_benchmarks[tag].run()
                logger.debug(benchmark_results1[tag].show())

        # Step #3
        patching_repos = patching.add_remote_repositories(
            self.env, settings.PATCHING_MIRRORS)
        if settings.PATCHING_MASTER_MIRRORS:
            patching_master_repos = patching.add_remote_repositories(
                self.env,
                settings.PATCHING_MASTER_MIRRORS,
                prefix_name='custom_master_repo')

        # Step #4
        slaves = self.fuel_web.client.list_cluster_nodes(cluster_id)
        for repo in patching_repos:
            patching.connect_slaves_to_repo(self.env, slaves, repo)
        if settings.PATCHING_MASTER_MIRRORS:
            with self.env.d_env.get_admin_remote() as remote:
                for repo in patching_master_repos:
                    install_pkg(remote, 'yum-utils')
                    patching.connect_admin_to_repo(self.env, repo)

        # Step #5
        if settings.LATE_ARTIFACTS_JOB_URL:
            data = urllib2.urlopen(settings.LATE_ARTIFACTS_JOB_URL +
                                   "/artifact/artifacts/artifacts.txt")
            for package in data:
                os.system("wget --directory-prefix"
                          " {0} {1}".format(settings.UPDATE_FUEL_PATH,
                                            package))
            self.env.admin_actions.upload_packages(
                local_packages_dir=settings.UPDATE_FUEL_PATH,
                centos_repo_path='/var/www/nailgun/centos/auxiliary',
                ubuntu_repo_path=settings.LOCAL_MIRROR_UBUNTU)
        if settings.REGENERATE_ENV_IMAGE:
            self.env.admin_actions.clean_generated_image(
                settings.OPENSTACK_RELEASE)

        # Step #6
        logger.info('Applying fix...')
        patching.apply_patches(self.env, target='environment', slaves=slaves)

        # Step #7
        logger.info('Verifying fix...')
        patching.verify_fix(self.env, target='environment', slaves=slaves)

        # Step #8
        # If OSTF fails (sometimes services aren't ready after
        # slaves nodes reboot) sleep 5 minutes and try again
        try:
            self.fuel_web.run_ostf(cluster_id=cluster_id)
        except AssertionError:
            time.sleep(300)
            self.fuel_web.run_ostf(cluster_id=cluster_id)

        # Step #9
        if settings.PATCHING_RUN_RALLY:
            benchmark_results2 = {}
            for tag in set(settings.RALLY_TAGS):
                benchmark_results2[tag] = rally_benchmarks[tag].run()
                logger.debug(benchmark_results2[tag].show())

            rally_benchmarks_passed = True

            for tag in set(settings.RALLY_TAGS):
                if not RallyResult.compare(benchmark_results1[tag],
                                           benchmark_results2[tag],
                                           deviation=0.2):
                    rally_benchmarks_passed = False

            assert_true(
                rally_benchmarks_passed,
                "Rally benchmarks show performance degradation "
                "after packages patching.")

        number_of_nodes = len(
            self.fuel_web.client.list_cluster_nodes(cluster_id))

        cluster_nodes = self.fuel_web.client.list_cluster_nodes(cluster_id)
        roles_list = [node['roles'] for node in cluster_nodes]
        unique_roles = []

        for role in roles_list:
            if not [
                    unique_role for unique_role in unique_roles
                    if set(role) == set(unique_role)
            ]:
                unique_roles.append(role)

        self.env.bootstrap_nodes(
            self.env.d_env.nodes().slaves[number_of_nodes:number_of_nodes + 1])

        for roles in unique_roles:
            if "mongo" in roles:
                continue

            node = {
                'slave-0{}'.format(number_of_nodes + 1):
                [role for role in roles]
            }
            logger.debug("Adding new node to the cluster: {0}".format(node))
            self.fuel_web.update_nodes(cluster_id, node)
            self.fuel_web.deploy_cluster_wait(cluster_id, check_services=False)
            self.fuel_web.verify_network(cluster_id)
            # sanity set isn't running due to LP1457515
            self.fuel_web.run_ostf(cluster_id=cluster_id,
                                   test_sets=['smoke', 'ha'])

            if "ceph-osd" in roles:
                with self.fuel_web.get_ssh_for_node(
                        'slave-0{}'.format(number_of_nodes +
                                           1)) as remote_ceph:
                    self.fuel_web.prepare_ceph_to_delete(remote_ceph)

            nailgun_node = self.fuel_web.update_nodes(cluster_id, node, False,
                                                      True)
            nodes = filter(lambda x: x["pending_deletion"] is True,
                           nailgun_node)
            self.fuel_web.deploy_cluster(cluster_id)
            wait(lambda: self.fuel_web.is_node_discovered(nodes[0]),
                 timeout=6 * 60)
            # sanity set isn't running due to LP1457515
            self.fuel_web.run_ostf(cluster_id=cluster_id,
                                   test_sets=['smoke', 'ha'])
Beispiel #2
0
    def patching_environment(self):
        """Apply patches on deployed environment

        Scenario:
        1. Revert snapshot of deployed environment
        2. Run Rally benchmark tests and store results
        3. Download patched packages on master node and make local repositories
        4. Add new local repositories on slave nodes
        5. Download late artifacts and clean generated images if needed
        6. Perform actions required to apply patches
        7. Verify that fix works
        8. Run OSTF
        9. Run Rally benchmark tests and compare results

        Duration 15m
        """

        # Step #1
        if not self.env.revert_snapshot(self.snapshot_name):
            raise PatchingTestException('Environment revert from snapshot "{0}'
                                        '" failed.'.format(self.snapshot_name))
        # Check that environment exists and it's ready for patching
        cluster_id = self.fuel_web.get_last_created_cluster()
        assert_is_not_none(cluster_id, 'Environment for patching not found.')

        # Step #2
        if settings.PATCHING_RUN_RALLY:
            rally_benchmarks = {}
            benchmark_results1 = {}
            for tag in set(settings.RALLY_TAGS):
                rally_benchmarks[tag] = RallyBenchmarkTest(
                    container_repo=settings.RALLY_DOCKER_REPO,
                    environment=self.env,
                    cluster_id=cluster_id,
                    test_type=tag
                )
                benchmark_results1[tag] = rally_benchmarks[tag].run()
                logger.debug(benchmark_results1[tag].show())

        # Step #3
        patching_repos = patching.add_remote_repositories(
            self.env, settings.PATCHING_MIRRORS)
        if settings.PATCHING_MASTER_MIRRORS:
            patching_master_repos = patching.add_remote_repositories(
                self.env, settings.PATCHING_MASTER_MIRRORS,
                prefix_name='custom_master_repo')

        # Step #4
        slaves = self.fuel_web.client.list_cluster_nodes(cluster_id)
        for repo in patching_repos:
            patching.connect_slaves_to_repo(self.env, slaves, repo)
        if settings.PATCHING_MASTER_MIRRORS:
            with self.env.d_env.get_admin_remote() as remote:
                for repo in patching_master_repos:
                    install_pkg(remote, 'yum-utils')
                    patching.connect_admin_to_repo(self.env, repo)

        # Step #5
        if settings.LATE_ARTIFACTS_JOB_URL:
            data = urlopen(settings.LATE_ARTIFACTS_JOB_URL +
                           "/artifact/artifacts/artifacts.txt")
            for package in data:
                os.system("wget --directory-prefix"
                          " {0} {1}".format(settings.UPDATE_FUEL_PATH,
                                            package))
            self.env.admin_actions.upload_packages(
                local_packages_dir=settings.UPDATE_FUEL_PATH,
                centos_repo_path='/var/www/nailgun/centos/auxiliary',
                ubuntu_repo_path=settings.LOCAL_MIRROR_UBUNTU)
        if settings.REGENERATE_ENV_IMAGE:
            self.env.admin_actions.clean_generated_image(
                settings.OPENSTACK_RELEASE)

        # Step #6
        logger.info('Applying fix...')
        patching.apply_patches(self.env, target='environment', slaves=slaves)

        # Step #7
        logger.info('Verifying fix...')
        patching.verify_fix(self.env, target='environment', slaves=slaves)

        # Step #8
        # If OSTF fails (sometimes services aren't ready after
        # slaves nodes reboot) sleep 5 minutes and try again
        try:
            self.fuel_web.run_ostf(cluster_id=cluster_id)
        except AssertionError:
            time.sleep(300)
            self.fuel_web.run_ostf(cluster_id=cluster_id)

        # Step #9
        if settings.PATCHING_RUN_RALLY:
            benchmark_results2 = {}
            for tag in set(settings.RALLY_TAGS):
                benchmark_results2[tag] = rally_benchmarks[tag].run()
                logger.debug(benchmark_results2[tag].show())

            rally_benchmarks_passed = True

            for tag in set(settings.RALLY_TAGS):
                if not RallyResult.compare(benchmark_results1[tag],
                                           benchmark_results2[tag],
                                           deviation=0.2):
                    rally_benchmarks_passed = False

            assert_true(rally_benchmarks_passed,
                        "Rally benchmarks show performance degradation "
                        "after packages patching.")

        number_of_nodes = len(self.fuel_web.client.list_cluster_nodes(
            cluster_id))

        cluster_nodes = self.fuel_web.client.list_cluster_nodes(cluster_id)
        roles_list = [node['roles'] for node in cluster_nodes]
        unique_roles = []

        for role in roles_list:
            if not [unique_role for unique_role in unique_roles
                    if set(role) == set(unique_role)]:
                unique_roles.append(role)

        self.env.bootstrap_nodes(self.env.d_env.nodes().slaves[
                                 number_of_nodes:number_of_nodes + 1])

        for roles in unique_roles:
            if "mongo" in roles:
                continue

            node = {'slave-0{}'.format(number_of_nodes + 1):
                    [role for role in roles]}
            logger.debug("Adding new node to the cluster: {0}".format(node))
            self.fuel_web.update_nodes(
                cluster_id, node)
            self.fuel_web.deploy_cluster_wait(cluster_id,
                                              check_services=False)
            self.fuel_web.verify_network(cluster_id)
            self.fuel_web.run_ostf(cluster_id=cluster_id,
                                   test_sets=['sanity', 'smoke', 'ha'])

            if "ceph-osd" in roles:
                with self.fuel_web.get_ssh_for_node(
                        'slave-0{}'.format(number_of_nodes + 1)
                ) as remote_ceph:
                    self.fuel_web.prepare_ceph_to_delete(remote_ceph)

            nailgun_node = self.fuel_web.update_nodes(
                cluster_id, node, False, True)
            nodes = [_node for _node in nailgun_node
                     if _node["pending_deletion"] is True]
            self.fuel_web.deploy_cluster(cluster_id)
            wait(
                lambda: self.fuel_web.is_node_discovered(nodes[0]),
                timeout=6 * 60)
            # sanity set isn't running due to LP1457515
            self.fuel_web.run_ostf(cluster_id=cluster_id,
                                   test_sets=['smoke', 'ha'])
Beispiel #3
0
    def patching_test(self):
        """Apply patches on deployed master

        Scenario:
        1. Download patched packages on master node and make local repositories
        2. Download late artifacts and clean generated images if needed
        3. Perform actions required to apply patches
        4. Verify that fix works
        5. Run OSTF
        6. Run network verification
        7. Reset and delete cluster
        8. Bootstrap 3 slaves

        Duration 30m
        """

        if not self.env.revert_snapshot(self.snapshot_name):
            raise PatchingTestException('Environment revert from snapshot "{0}'
                                        '" failed.'.format(self.snapshot_name))

        # Step #1
        with self.env.d_env.get_admin_remote() as remote:
            install_pkg(remote, 'yum-utils')
        patching_repos = patching.add_remote_repositories(
            self.env, settings.PATCHING_MASTER_MIRRORS)

        for repo in patching_repos:
            patching.connect_admin_to_repo(self.env, repo)

        # Step #2
        if settings.LATE_ARTIFACTS_JOB_URL:
            data = urllib2.urlopen(settings.LATE_ARTIFACTS_JOB_URL +
                                   "/artifact/artifacts/artifacts.txt")
            for package in data:
                os.system("wget --directory-prefix"
                          " {0} {1}".format(settings.UPDATE_FUEL_PATH,
                                            package))
            self.env.admin_actions.upload_packages(
                local_packages_dir=settings.UPDATE_FUEL_PATH,
                centos_repo_path='/var/www/nailgun/centos/auxiliary',
                ubuntu_repo_path=settings.LOCAL_MIRROR_UBUNTU)
        if settings.REGENERATE_ENV_IMAGE:
            self.env.admin_actions.clean_generated_image(
                settings.OPENSTACK_RELEASE)

        # Step #3
        logger.info('Applying fix...')
        patching.apply_patches(self.env, target='master')

        # Step #4
        logger.info('Verifying fix...')
        patching.verify_fix(self.env, target='master')

        # Step #5
        active_nodes = []
        for node in self.env.d_env.nodes().slaves:
            if node.driver.node_active(node):
                active_nodes.append(node)
        logger.debug('active nodes are {}'.format(active_nodes))
        cluster_id = self.fuel_web.get_last_created_cluster()
        if self.fuel_web.get_last_created_cluster():
            number_of_nodes = len(
                self.fuel_web.client.list_cluster_nodes(cluster_id))
            self.fuel_web.run_ostf(cluster_id=cluster_id)
            if number_of_nodes > 1:
                self.fuel_web.verify_network(cluster_id)

            cluster_nodes = self.fuel_web.client.list_cluster_nodes(cluster_id)
            roles_list = [node['roles'] for node in cluster_nodes]
            unique_roles = []

            for role in roles_list:
                if not [
                        unique_role for unique_role in unique_roles
                        if set(role) == set(unique_role)
                ]:
                    unique_roles.append(role)

            self.env.bootstrap_nodes(
                self.env.d_env.nodes().slaves[number_of_nodes:number_of_nodes +
                                              1])

            for roles in unique_roles:
                if "mongo" in roles:
                    continue
                node = {
                    'slave-0{}'.format(number_of_nodes + 1):
                    [role for role in roles]
                }
                logger.debug("Adding new node to"
                             " the cluster: {0}".format(node))
                self.fuel_web.update_nodes(cluster_id, node)
                self.fuel_web.deploy_cluster_wait(cluster_id,
                                                  check_services=False)
                self.fuel_web.verify_network(cluster_id)
                # sanity set isn't running due to LP1457515
                self.fuel_web.run_ostf(cluster_id=cluster_id,
                                       test_sets=['smoke', 'ha'])

                if "ceph-osd" in roles:
                    with self.fuel_web.get_ssh_for_node(
                            'slave-0{}'.format(number_of_nodes + 1)) as remote:
                        self.fuel_web.prepare_ceph_to_delete(remote)
                nailgun_node = self.fuel_web.update_nodes(
                    cluster_id, node, False, True)
                nodes = filter(lambda x: x["pending_deletion"] is True,
                               nailgun_node)
                self.fuel_web.deploy_cluster(cluster_id)
                wait(lambda: self.fuel_web.is_node_discovered(nodes[0]),
                     timeout=6 * 60)
                # sanity set isn't running due to LP1457515
                self.fuel_web.run_ostf(cluster_id=cluster_id,
                                       test_sets=['smoke', 'ha'])

            active_nodes = []
            for node in self.env.d_env.nodes().slaves:
                if node.driver.node_active(node):
                    active_nodes.append(node)
            logger.debug('active nodes are {}'.format(active_nodes))

            self.fuel_web.stop_reset_env_wait(cluster_id)
            self.fuel_web.wait_nodes_get_online_state(active_nodes,
                                                      timeout=10 * 60)
            self.fuel_web.client.delete_cluster(cluster_id)
            try:
                wait((lambda: len(self.fuel_web.client.list_nodes()) ==
                      number_of_nodes),
                     timeout=5 * 60)
            except TimeoutError:
                assert_true(
                    len(self.fuel_web.client.list_nodes()) == number_of_nodes,
                    'Nodes are not discovered in timeout 5 *60')
        self.env.bootstrap_nodes(self.env.d_env.nodes().slaves[:3])
Beispiel #4
0
    def patching_test(self):
        """Apply patches on deployed master

        Scenario:
        1. Download patched packages on master node and make local repositories
        2. Download late artifacts and clean generated images if needed
        3. Perform actions required to apply patches
        4. Verify that fix works
        5. Run OSTF
        6. Run network verification
        7. Reset and delete cluster
        8. Bootstrap 3 slaves

        Duration 30m
        """

        if not self.env.revert_snapshot(self.snapshot_name):
            raise PatchingTestException('Environment revert from snapshot "{0}'
                                        '" failed.'.format(self.snapshot_name))

        # Step #1
        with self.env.d_env.get_admin_remote() as remote:
            install_pkg(remote, 'yum-utils')
        patching_repos = patching.add_remote_repositories(
            self.env, settings.PATCHING_MASTER_MIRRORS)

        for repo in patching_repos:
            patching.connect_admin_to_repo(self.env, repo)

        # Step #2
        if settings.LATE_ARTIFACTS_JOB_URL:
            data = urlopen(
                settings.LATE_ARTIFACTS_JOB_URL +
                "/artifact/artifacts/artifacts.txt")
            for package in data:
                os.system("wget --directory-prefix"
                          " {0} {1}".format(settings.UPDATE_FUEL_PATH,
                                            package))
            self.env.admin_actions.upload_packages(
                local_packages_dir=settings.UPDATE_FUEL_PATH,
                centos_repo_path='/var/www/nailgun/centos/auxiliary',
                ubuntu_repo_path=settings.LOCAL_MIRROR_UBUNTU)
        if settings.REGENERATE_ENV_IMAGE:
            self.env.admin_actions.clean_generated_image(
                settings.OPENSTACK_RELEASE)

        # Step #3
        logger.info('Applying fix...')
        patching.apply_patches(self.env, target='master')

        # Step #4
        logger.info('Verifying fix...')
        patching.verify_fix(self.env, target='master')

        # Step #5
        active_nodes = []
        for node in self.env.d_env.nodes().slaves:
            if node.driver.node_active(node):
                active_nodes.append(node)
        logger.debug('active nodes are {}'.format(active_nodes))
        cluster_id = self.fuel_web.get_last_created_cluster()
        if self.fuel_web.get_last_created_cluster():
            number_of_nodes = len(self.fuel_web.client.list_cluster_nodes(
                cluster_id))
            self.fuel_web.run_ostf(cluster_id=cluster_id)
            if number_of_nodes > 1:
                self.fuel_web.verify_network(cluster_id)

            cluster_nodes = self.fuel_web.client.list_cluster_nodes(cluster_id)
            roles_list = [node['roles'] for node in cluster_nodes]
            unique_roles = []

            for role in roles_list:
                if not [unique_role for unique_role in unique_roles
                        if set(role) == set(unique_role)]:
                    unique_roles.append(role)

            self.env.bootstrap_nodes(self.env.d_env.nodes().slaves[
                                     number_of_nodes:number_of_nodes + 1])

            for roles in unique_roles:
                if "mongo" in roles:
                    continue
                node = {'slave-0{}'.format(number_of_nodes + 1):
                        [role for role in roles]}
                logger.debug("Adding new node to"
                             " the cluster: {0}".format(node))
                self.fuel_web.update_nodes(
                    cluster_id, node)
                self.fuel_web.deploy_cluster_wait(cluster_id,
                                                  check_services=False)
                self.fuel_web.verify_network(cluster_id)

                self.fuel_web.run_ostf(cluster_id=cluster_id,
                                       test_sets=['sanity', 'smoke', 'ha'])

                if "ceph-osd" in roles:
                    with self.fuel_web.get_ssh_for_node(
                            'slave-0{}'.format(number_of_nodes + 1)
                    ) as remote:
                        self.fuel_web.prepare_ceph_to_delete(remote)
                nailgun_node = self.fuel_web.update_nodes(
                    cluster_id, node, False, True)
                nodes = [_node for _node in nailgun_node
                         if _node["pending_deletion"] is True]
                self.fuel_web.deploy_cluster(cluster_id)
                wait(
                    lambda: self.fuel_web.is_node_discovered(nodes[0]),
                    timeout=6 * 60)
                # sanity set isn't running due to LP1457515
                self.fuel_web.run_ostf(cluster_id=cluster_id,
                                       test_sets=['smoke', 'ha'])

            active_nodes = []
            for node in self.env.d_env.nodes().slaves:
                if node.driver.node_active(node):
                    active_nodes.append(node)
            logger.debug('active nodes are {}'.format(active_nodes))

            self.fuel_web.stop_reset_env_wait(cluster_id)
            self.fuel_web.wait_nodes_get_online_state(
                active_nodes, timeout=10 * 60)
            self.fuel_web.client.delete_cluster(cluster_id)
            try:
                wait((lambda: len(
                    self.fuel_web.client.list_nodes()) == number_of_nodes),
                    timeout=5 * 60)
            except TimeoutError:
                assert_true(len(
                    self.fuel_web.client.list_nodes()) == number_of_nodes,
                    'Nodes are not discovered in timeout 5 *60')
        self.env.bootstrap_nodes(self.env.d_env.nodes().slaves[:3])