diff options
Diffstat (limited to 'roles/openshift_health_checker')
10 files changed, 454 insertions, 161 deletions
diff --git a/roles/openshift_health_checker/action_plugins/openshift_health_check.py b/roles/openshift_health_checker/action_plugins/openshift_health_check.py index 03c40b78b..a62e4331e 100644 --- a/roles/openshift_health_checker/action_plugins/openshift_health_check.py +++ b/roles/openshift_health_checker/action_plugins/openshift_health_check.py @@ -25,9 +25,11 @@ class ActionModule(ActionBase): def run(self, tmp=None, task_vars=None): result = super(ActionModule, self).run(tmp, task_vars) + task_vars = task_vars or {} - if task_vars is None: - task_vars = {} + # vars are not supportably available in the callback plugin, + # so record any it will need in the result. + result['playbook_context'] = task_vars.get('r_openshift_health_checker_playbook_context') if "openshift" not in task_vars: result["failed"] = True @@ -46,19 +48,27 @@ class ActionModule(ActionBase): result["checks"] = check_results = {} + user_disabled_checks = [ + check.strip() + for check in task_vars.get("openshift_disable_check", "").split(",") + ] + for check_name in resolved_checks: display.banner("CHECK [{} : {}]".format(check_name, task_vars["ansible_host"])) check = known_checks[check_name] - if check.is_active(task_vars): + if not check.is_active(task_vars): + r = dict(skipped=True, skipped_reason="Not active for this host") + elif check_name in user_disabled_checks: + r = dict(skipped=True, skipped_reason="Disabled by user request") + else: try: r = check.run(tmp, task_vars) except OpenShiftCheckException as e: - r = {} - r["failed"] = True - r["msg"] = str(e) - else: - r = {"skipped": True} + r = dict( + failed=True, + msg=str(e), + ) check_results[check_name] = r diff --git a/roles/openshift_health_checker/callback_plugins/zz_failure_summary.py b/roles/openshift_health_checker/callback_plugins/zz_failure_summary.py index 7bce7f107..64c29a8d9 100644 --- a/roles/openshift_health_checker/callback_plugins/zz_failure_summary.py +++ b/roles/openshift_health_checker/callback_plugins/zz_failure_summary.py @@ -2,6 +2,12 @@ Ansible callback plugin. ''' +# Reason: In several locations below we disable pylint protected-access +# for Ansible objects that do not give us any public way +# to access the full details we need to report check failures. +# Status: disabled permanently or until Ansible object has a public API. +# This does leave the code more likely to be broken by future Ansible changes. + from pprint import pformat from ansible.plugins.callback import CallbackBase @@ -20,38 +26,37 @@ class CallbackModule(CallbackBase): CALLBACK_TYPE = 'aggregate' CALLBACK_NAME = 'failure_summary' CALLBACK_NEEDS_WHITELIST = False + _playbook_file = None def __init__(self): super(CallbackModule, self).__init__() self.__failures = [] + def v2_playbook_on_start(self, playbook): + super(CallbackModule, self).v2_playbook_on_start(playbook) + # re: playbook attrs see top comment # pylint: disable=protected-access + self._playbook_file = playbook._file_name + def v2_runner_on_failed(self, result, ignore_errors=False): super(CallbackModule, self).v2_runner_on_failed(result, ignore_errors) self.__failures.append(dict(result=result, ignore_errors=ignore_errors)) def v2_playbook_on_stats(self, stats): super(CallbackModule, self).v2_playbook_on_stats(stats) - # TODO: update condition to consider a host var or env var to - # enable/disable the summary, so that we can control the output from a - # play. if self.__failures: - self._print_failure_summary() + self._print_failure_details(self.__failures) - def _print_failure_summary(self): - '''Print a summary of failed tasks (including ignored failures).''' + def _print_failure_details(self, failures): + '''Print a summary of failed tasks or checks.''' self._display.display(u'\nFailure summary:\n') - # TODO: group failures by host or by task. If grouped by host, it is - # easy to see all problems of a given host. If grouped by task, it is - # easy to see what hosts needs the same fix. - - width = len(str(len(self.__failures))) + width = len(str(len(failures))) initial_indent_format = u' {{:>{width}}}. '.format(width=width) initial_indent_len = len(initial_indent_format.format(0)) subsequent_indent = u' ' * initial_indent_len subsequent_extra_indent = u' ' * (initial_indent_len + 10) - for i, failure in enumerate(self.__failures, 1): + for i, failure in enumerate(failures, 1): entries = _format_failure(failure) self._display.display(u'\n{}{}'.format(initial_indent_format.format(i), entries[0])) for entry in entries[1:]: @@ -59,11 +64,52 @@ class CallbackModule(CallbackBase): indented = u'{}{}'.format(subsequent_indent, entry) self._display.display(indented) - -# Reason: disable pylint protected-access because we need to access _* -# attributes of a task result to implement this method. -# Status: permanently disabled unless Ansible's API changes. -# pylint: disable=protected-access + failed_checks = set() + playbook_context = None + # re: result attrs see top comment # pylint: disable=protected-access + for failure in failures: + # get context from check task result since callback plugins cannot access task vars + playbook_context = playbook_context or failure['result']._result.get('playbook_context') + failed_checks.update( + name + for name, result in failure['result']._result.get('checks', {}).items() + if result.get('failed') + ) + if failed_checks: + self._print_check_failure_summary(failed_checks, playbook_context) + + def _print_check_failure_summary(self, failed_checks, context): + checks = ','.join(sorted(failed_checks)) + # NOTE: context is not set if all failures occurred prior to checks task + summary = ( + '\n' + 'The execution of "{playbook}"\n' + 'includes checks designed to fail early if the requirements\n' + 'of the playbook are not met. One or more of these checks\n' + 'failed. To disregard these results, you may choose to\n' + 'disable failing checks by setting an Ansible variable:\n\n' + ' openshift_disable_check={checks}\n\n' + 'Failing check names are shown in the failure details above.\n' + 'Some checks may be configurable by variables if your requirements\n' + 'are different from the defaults; consult check documentation.\n' + 'Variables can be set in the inventory or passed on the\n' + 'command line using the -e flag to ansible-playbook.\n' + ).format(playbook=self._playbook_file, checks=checks) + if context in ['pre-install', 'health']: + summary = ( + '\n' + 'You may choose to configure or disable failing checks by\n' + 'setting Ansible variables. To disable those above:\n\n' + ' openshift_disable_check={checks}\n\n' + 'Consult check documentation for configurable variables.\n' + 'Variables can be set in the inventory or passed on the\n' + 'command line using the -e flag to ansible-playbook.\n' + ).format(checks=checks) + # other expected contexts: install, upgrade + self._display.display(summary) + + +# re: result attrs see top comment # pylint: disable=protected-access def _format_failure(failure): '''Return a list of pretty-formatted text entries describing a failure, including relevant information about it. Expect that the list of text entries will be joined @@ -100,11 +146,8 @@ def _format_failed_checks(checks): return stringc(pformat(checks), C.COLOR_ERROR) -# Reason: disable pylint protected-access because we need to access _* -# attributes of obj to implement this function. -# This is inspired by ansible.playbook.base.Base.dump_me. -# Status: permanently disabled unless Ansible's API changes. -# pylint: disable=protected-access +# This is inspired by ansible.playbook.base.Base.dump_me. +# re: play/task/block attrs see top comment # pylint: disable=protected-access def _get_play(obj): '''Given a task or block, recursively tries to find its parent play.''' if hasattr(obj, '_play'): diff --git a/roles/openshift_health_checker/meta/main.yml b/roles/openshift_health_checker/meta/main.yml index cd9b55902..4d141974c 100644 --- a/roles/openshift_health_checker/meta/main.yml +++ b/roles/openshift_health_checker/meta/main.yml @@ -2,3 +2,4 @@ dependencies: - role: openshift_facts - role: openshift_repos + - role: openshift_version diff --git a/roles/openshift_health_checker/openshift_checks/disk_availability.py b/roles/openshift_health_checker/openshift_checks/disk_availability.py index c2792a0fe..962148cb8 100644 --- a/roles/openshift_health_checker/openshift_checks/disk_availability.py +++ b/roles/openshift_health_checker/openshift_checks/disk_availability.py @@ -27,10 +27,12 @@ class DiskAvailability(NotContainerizedMixin, OpenShiftCheck): def run(self, tmp, task_vars): group_names = get_var(task_vars, "group_names") ansible_mounts = get_var(task_vars, "ansible_mounts") - - min_free_bytes = max(self.recommended_disk_space_bytes.get(name, 0) for name in group_names) free_bytes = self.openshift_available_disk(ansible_mounts) + recommended_min = max(self.recommended_disk_space_bytes.get(name, 0) for name in group_names) + configured_min = int(get_var(task_vars, "openshift_check_min_host_disk_gb", default=0)) * 10**9 + min_free_bytes = configured_min or recommended_min + if free_bytes < min_free_bytes: return { 'failed': True, diff --git a/roles/openshift_health_checker/openshift_checks/docker_image_availability.py b/roles/openshift_health_checker/openshift_checks/docker_image_availability.py index cce289b95..4588ed634 100644 --- a/roles/openshift_health_checker/openshift_checks/docker_image_availability.py +++ b/roles/openshift_health_checker/openshift_checks/docker_image_availability.py @@ -13,41 +13,55 @@ class DockerImageAvailability(OpenShiftCheck): name = "docker_image_availability" tags = ["preflight"] - skopeo_image = "openshift/openshift-ansible" + dependencies = ["skopeo", "python-docker-py"] - # FIXME(juanvallejo): we should consider other possible values of - # `deployment_type` (the key here). See - # https://github.com/openshift/openshift-ansible/blob/8e26f8c/roles/openshift_repos/vars/main.yml#L7 - docker_image_base = { + deployment_image_info = { "origin": { - "repo": "openshift", - "image": "origin", + "namespace": "openshift", + "name": "origin", }, "openshift-enterprise": { - "repo": "openshift3", - "image": "ose", + "namespace": "openshift3", + "name": "ose", }, } - def run(self, tmp, task_vars): - required_images = self.required_images(task_vars) - missing_images = set(required_images) - set(self.local_images(required_images, task_vars)) + @classmethod + def is_active(cls, task_vars): + """Skip hosts with unsupported deployment types.""" + deployment_type = get_var(task_vars, "openshift_deployment_type") + has_valid_deployment_type = deployment_type in cls.deployment_image_info - # exit early if all images were found locally - if not missing_images: - return {"changed": False} + return super(DockerImageAvailability, cls).is_active(task_vars) and has_valid_deployment_type - msg, failed, changed = self.update_skopeo_image(task_vars) + def run(self, tmp, task_vars): + msg, failed, changed = self.ensure_dependencies(task_vars) # exit early if Skopeo update fails if failed: + if "No package matching" in msg: + msg = "Ensure that all required dependencies can be installed via `yum`.\n" return { "failed": True, "changed": changed, - "msg": "Failed to update Skopeo image ({img_name}). {msg}".format(img_name=self.skopeo_image, msg=msg), + "msg": ( + "Unable to update or install required dependency packages on this host;\n" + "These are required in order to check Docker image availability:" + "\n {deps}\n{msg}" + ).format(deps=',\n '.join(self.dependencies), msg=msg), } + required_images = self.required_images(task_vars) + missing_images = set(required_images) - set(self.local_images(required_images, task_vars)) + + # exit early if all images were found locally + if not missing_images: + return {"changed": changed} + registries = self.known_docker_registries(task_vars) + if not registries: + return {"failed": True, "msg": "Unable to retrieve any docker registries.", "changed": changed} + available_images = self.available_images(missing_images, registries, task_vars) unavailable_images = set(missing_images) - set(available_images) @@ -55,44 +69,60 @@ class DockerImageAvailability(OpenShiftCheck): return { "failed": True, "msg": ( - "One or more required images are not available: {}.\n" + "One or more required Docker images are not available:\n {}\n" "Configured registries: {}" - ).format(", ".join(sorted(unavailable_images)), ", ".join(registries)), + ).format(",\n ".join(sorted(unavailable_images)), ", ".join(registries)), "changed": changed, } return {"changed": changed} def required_images(self, task_vars): - deployment_type = get_var(task_vars, "deployment_type") - # FIXME(juanvallejo): we should handle gracefully with a proper error - # message when given an unexpected value for `deployment_type`. - image_base_name = self.docker_image_base[deployment_type] - - openshift_release = get_var(task_vars, "openshift_release") - # FIXME(juanvallejo): this variable is not required when the - # installation is non-containerized. The example inventories have it - # commented out. We should handle gracefully and with a proper error - # message when this variable is required and not set. - openshift_image_tag = get_var(task_vars, "openshift_image_tag") + deployment_type = get_var(task_vars, "openshift_deployment_type") + image_info = self.deployment_image_info[deployment_type] + openshift_release = get_var(task_vars, "openshift_release", default="latest") + openshift_image_tag = get_var(task_vars, "openshift_image_tag") is_containerized = get_var(task_vars, "openshift", "common", "is_containerized") - if is_containerized: - images = set(self.containerized_docker_images(image_base_name, openshift_release)) - else: - images = set(self.rpm_docker_images(image_base_name, openshift_release)) + images = set(self.required_docker_images( + image_info["namespace"], + image_info["name"], + ["registry-console"] if "enterprise" in deployment_type else [], # include enterprise-only image names + openshift_release, + is_containerized, + )) # append images with qualified image tags to our list of required images. # these are images with a (v0.0.0.0) tag, rather than a standard release # format tag (v0.0). We want to check this set in both containerized and # non-containerized installations. images.update( - self.qualified_docker_images(self.image_from_base_name(image_base_name), "v" + openshift_image_tag) + self.required_qualified_docker_images( + image_info["namespace"], + image_info["name"], + openshift_image_tag, + ), ) return images + @staticmethod + def required_docker_images(namespace, name, additional_image_names, version, is_containerized): + if is_containerized: + return ["{}/{}:{}".format(namespace, name, version)] if name else [] + + # include additional non-containerized images specific to the current deployment type + return ["{}/{}:{}".format(namespace, img_name, version) for img_name in additional_image_names] + + @staticmethod + def required_qualified_docker_images(namespace, name, version): + # pylint: disable=invalid-name + return [ + "{}/{}-{}:{}".format(namespace, name, suffix, version) + for suffix in ["haproxy-router", "docker-registry", "deployer", "pod"] + ] + def local_images(self, images, task_vars): """Filter a list of images and return those available locally.""" return [ @@ -107,31 +137,26 @@ class DockerImageAvailability(OpenShiftCheck): return bool(result.get("images", [])) - def known_docker_registries(self, task_vars): - result = self.module_executor("docker_info", {}, task_vars) + @staticmethod + def known_docker_registries(task_vars): + docker_facts = get_var(task_vars, "openshift", "docker") + regs = set(docker_facts["additional_registries"]) - if result.get("failed", False): - return [] + deployment_type = get_var(task_vars, "openshift_deployment_type") + if deployment_type == "origin": + regs.update(["docker.io"]) + elif "enterprise" in deployment_type: + regs.update(["registry.access.redhat.com"]) - # FIXME(juanvallejo): wrong default type, result["info"] is expected to - # contain a dictionary (see how we call `docker_info.get` below). - docker_info = result.get("info", "") - return [registry.get("Name", "") for registry in docker_info.get("Registries", {})] + return list(regs) def available_images(self, images, registries, task_vars): """Inspect existing images using Skopeo and return all images successfully inspected.""" return [ image for image in images - if self.is_image_available(image, registries, task_vars) + if any(self.is_available_skopeo_image(image, registry, task_vars) for registry in registries) ] - def is_image_available(self, image, registries, task_vars): - for registry in registries: - if self.is_available_skopeo_image(image, registry, task_vars): - return True - - return False - def is_available_skopeo_image(self, image, registry, task_vars): """Uses Skopeo to determine if required image exists in a given registry.""" @@ -140,40 +165,15 @@ class DockerImageAvailability(OpenShiftCheck): image=image, ) - args = { - "name": "skopeo_inspect", - "image": self.skopeo_image, - "command": cmd_str, - "detach": False, - "cleanup": True, - } - result = self.module_executor("docker_container", args, task_vars) - return result.get("failed", False) - - def containerized_docker_images(self, base_name, version): - return [ - "{image}:{version}".format(image=self.image_from_base_name(base_name), version=version) - ] + args = {"_raw_params": cmd_str} + result = self.module_executor("command", args, task_vars) + return not result.get("failed", False) and result.get("rc", 0) == 0 - @staticmethod - def rpm_docker_images(base, version): - return [ - "{image_repo}/registry-console:{version}".format(image_repo=base["repo"], version=version) - ] + # ensures that the skopeo and python-docker-py packages exist + # check is skipped on atomic installations + def ensure_dependencies(self, task_vars): + if get_var(task_vars, "openshift", "common", "is_atomic"): + return "", False, False - @staticmethod - def qualified_docker_images(image_name, version): - return [ - "{}-{}:{}".format(image_name, component, version) - for component in "haproxy-router docker-registry deployer pod".split() - ] - - @staticmethod - def image_from_base_name(base): - return "".join([base["repo"], "/", base["image"]]) - - # ensures that the skopeo docker image exists, and updates it - # with latest if image was already present locally. - def update_skopeo_image(self, task_vars): - result = self.module_executor("docker_image", {"name": self.skopeo_image}, task_vars) - return result.get("msg", ""), result.get("failed", False), result.get("changed", False) + result = self.module_executor("yum", {"name": self.dependencies, "state": "latest"}, task_vars) + return result.get("msg", ""), result.get("failed", False) or result.get("rc", 0) != 0, result.get("changed") diff --git a/roles/openshift_health_checker/openshift_checks/memory_availability.py b/roles/openshift_health_checker/openshift_checks/memory_availability.py index 28805dc37..f4e31065f 100644 --- a/roles/openshift_health_checker/openshift_checks/memory_availability.py +++ b/roles/openshift_health_checker/openshift_checks/memory_availability.py @@ -1,6 +1,9 @@ # pylint: disable=missing-docstring from openshift_checks import OpenShiftCheck, get_var +MIB = 2**20 +GIB = 2**30 + class MemoryAvailability(OpenShiftCheck): """Check that recommended memory is available.""" @@ -11,10 +14,12 @@ class MemoryAvailability(OpenShiftCheck): # Values taken from the official installation documentation: # https://docs.openshift.org/latest/install_config/install/prerequisites.html#system-requirements recommended_memory_bytes = { - "masters": 16 * 10**9, - "nodes": 8 * 10**9, - "etcd": 20 * 10**9, + "masters": 16 * GIB, + "nodes": 8 * GIB, + "etcd": 8 * GIB, } + # https://access.redhat.com/solutions/3006511 physical RAM is partly reserved from memtotal + memtotal_adjustment = 1 * GIB @classmethod def is_active(cls, task_vars): @@ -25,19 +30,21 @@ class MemoryAvailability(OpenShiftCheck): def run(self, tmp, task_vars): group_names = get_var(task_vars, "group_names") - total_memory_bytes = get_var(task_vars, "ansible_memtotal_mb") * 10**6 + total_memory_bytes = get_var(task_vars, "ansible_memtotal_mb") * MIB - min_memory_bytes = max(self.recommended_memory_bytes.get(name, 0) for name in group_names) + recommended_min = max(self.recommended_memory_bytes.get(name, 0) for name in group_names) + configured_min = float(get_var(task_vars, "openshift_check_min_host_memory_gb", default=0)) * GIB + min_memory_bytes = configured_min or recommended_min - if total_memory_bytes < min_memory_bytes: + if total_memory_bytes + self.memtotal_adjustment < min_memory_bytes: return { 'failed': True, 'msg': ( - 'Available memory ({available:.1f} GB) ' - 'below recommended value ({recommended:.1f} GB)' + 'Available memory ({available:.1f} GiB) is too far ' + 'below recommended value ({recommended:.1f} GiB)' ).format( - available=float(total_memory_bytes) / 10**9, - recommended=float(min_memory_bytes) / 10**9, + available=float(total_memory_bytes) / GIB, + recommended=float(min_memory_bytes) / GIB, ), } diff --git a/roles/openshift_health_checker/test/action_plugin_test.py b/roles/openshift_health_checker/test/action_plugin_test.py index 2693ae37b..6ebf0ebb2 100644 --- a/roles/openshift_health_checker/test/action_plugin_test.py +++ b/roles/openshift_health_checker/test/action_plugin_test.py @@ -67,6 +67,7 @@ def changed(result): return result.get('changed', False) +# tests whether task is skipped, not individual checks def skipped(result): return result.get('skipped', False) @@ -101,7 +102,20 @@ def test_action_plugin_skip_non_active_checks(plugin, task_vars, monkeypatch): result = plugin.run(tmp=None, task_vars=task_vars) - assert result['checks']['fake_check'] == {'skipped': True} + assert result['checks']['fake_check'] == dict(skipped=True, skipped_reason="Not active for this host") + assert not failed(result) + assert not changed(result) + assert not skipped(result) + + +def test_action_plugin_skip_disabled_checks(plugin, task_vars, monkeypatch): + checks = [fake_check('fake_check', is_active=True)] + monkeypatch.setattr('openshift_checks.OpenShiftCheck.subclasses', classmethod(lambda cls: checks)) + + task_vars['openshift_disable_check'] = 'fake_check' + result = plugin.run(tmp=None, task_vars=task_vars) + + assert result['checks']['fake_check'] == dict(skipped=True, skipped_reason="Disabled by user request") assert not failed(result) assert not changed(result) assert not skipped(result) diff --git a/roles/openshift_health_checker/test/disk_availability_test.py b/roles/openshift_health_checker/test/disk_availability_test.py index 970b474d7..b353fa610 100644 --- a/roles/openshift_health_checker/test/disk_availability_test.py +++ b/roles/openshift_health_checker/test/disk_availability_test.py @@ -42,9 +42,10 @@ def test_cannot_determine_available_disk(ansible_mounts, extra_words): assert word in str(excinfo.value) -@pytest.mark.parametrize('group_names,ansible_mounts', [ +@pytest.mark.parametrize('group_names,configured_min,ansible_mounts', [ ( ['masters'], + 0, [{ 'mount': '/', 'size_available': 40 * 10**9 + 1, @@ -52,6 +53,7 @@ def test_cannot_determine_available_disk(ansible_mounts, extra_words): ), ( ['nodes'], + 0, [{ 'mount': '/', 'size_available': 15 * 10**9 + 1, @@ -59,6 +61,7 @@ def test_cannot_determine_available_disk(ansible_mounts, extra_words): ), ( ['etcd'], + 0, [{ 'mount': '/', 'size_available': 20 * 10**9 + 1, @@ -66,6 +69,15 @@ def test_cannot_determine_available_disk(ansible_mounts, extra_words): ), ( ['etcd'], + 1, # configure lower threshold + [{ + 'mount': '/', + 'size_available': 1 * 10**9 + 1, # way smaller than recommended + }], + ), + ( + ['etcd'], + 0, [{ # not enough space on / ... 'mount': '/', @@ -77,9 +89,10 @@ def test_cannot_determine_available_disk(ansible_mounts, extra_words): }], ), ]) -def test_succeeds_with_recommended_disk_space(group_names, ansible_mounts): +def test_succeeds_with_recommended_disk_space(group_names, configured_min, ansible_mounts): task_vars = dict( group_names=group_names, + openshift_check_min_host_disk_gb=configured_min, ansible_mounts=ansible_mounts, ) @@ -89,9 +102,10 @@ def test_succeeds_with_recommended_disk_space(group_names, ansible_mounts): assert not result.get('failed', False) -@pytest.mark.parametrize('group_names,ansible_mounts,extra_words', [ +@pytest.mark.parametrize('group_names,configured_min,ansible_mounts,extra_words', [ ( ['masters'], + 0, [{ 'mount': '/', 'size_available': 1, @@ -99,7 +113,17 @@ def test_succeeds_with_recommended_disk_space(group_names, ansible_mounts): ['0.0 GB'], ), ( + ['masters'], + 100, # set a higher threshold + [{ + 'mount': '/', + 'size_available': 50 * 10**9, # would normally be enough... + }], + ['100.0 GB'], + ), + ( ['nodes'], + 0, [{ 'mount': '/', 'size_available': 1 * 10**9, @@ -108,6 +132,7 @@ def test_succeeds_with_recommended_disk_space(group_names, ansible_mounts): ), ( ['etcd'], + 0, [{ 'mount': '/', 'size_available': 1, @@ -116,6 +141,7 @@ def test_succeeds_with_recommended_disk_space(group_names, ansible_mounts): ), ( ['nodes', 'masters'], + 0, [{ 'mount': '/', # enough space for a node, not enough for a master @@ -125,6 +151,7 @@ def test_succeeds_with_recommended_disk_space(group_names, ansible_mounts): ), ( ['etcd'], + 0, [{ # enough space on / ... 'mount': '/', @@ -137,9 +164,10 @@ def test_succeeds_with_recommended_disk_space(group_names, ansible_mounts): ['0.0 GB'], ), ]) -def test_fails_with_insufficient_disk_space(group_names, ansible_mounts, extra_words): +def test_fails_with_insufficient_disk_space(group_names, configured_min, ansible_mounts, extra_words): task_vars = dict( group_names=group_names, + openshift_check_min_host_disk_gb=configured_min, ansible_mounts=ansible_mounts, ) diff --git a/roles/openshift_health_checker/test/docker_image_availability_test.py b/roles/openshift_health_checker/test/docker_image_availability_test.py index 2a9c32f77..0379cafb5 100644 --- a/roles/openshift_health_checker/test/docker_image_availability_test.py +++ b/roles/openshift_health_checker/test/docker_image_availability_test.py @@ -3,26 +3,176 @@ import pytest from openshift_checks.docker_image_availability import DockerImageAvailability -@pytest.mark.xfail(strict=True) # TODO: remove this once this test is fully implemented. -@pytest.mark.parametrize('task_vars,expected_result', [ - ( - dict( - openshift=dict(common=dict( +@pytest.mark.parametrize('deployment_type,is_active', [ + ("origin", True), + ("openshift-enterprise", True), + ("enterprise", False), + ("online", False), + ("invalid", False), + ("", False), +]) +def test_is_active(deployment_type, is_active): + task_vars = dict( + openshift_deployment_type=deployment_type, + ) + assert DockerImageAvailability.is_active(task_vars=task_vars) == is_active + + +@pytest.mark.parametrize("is_containerized,is_atomic", [ + (True, True), + (False, False), + (True, False), + (False, True), +]) +def test_all_images_available_locally(is_containerized, is_atomic): + def execute_module(module_name, args, task_vars): + if module_name == "yum": + return {"changed": True} + + assert module_name == "docker_image_facts" + assert 'name' in args + assert args['name'] + return { + 'images': [args['name']], + } + + result = DockerImageAvailability(execute_module=execute_module).run(tmp=None, task_vars=dict( + openshift=dict( + common=dict( + service_type='origin', + is_containerized=is_containerized, + is_atomic=is_atomic, + ), + docker=dict(additional_registries=["docker.io"]), + ), + openshift_deployment_type='origin', + openshift_release='v3.4', + openshift_image_tag='3.4', + )) + + assert not result.get('failed', False) + + +@pytest.mark.parametrize("available_locally", [ + False, + True, +]) +def test_all_images_available_remotely(available_locally): + def execute_module(module_name, args, task_vars): + if module_name == 'docker_image_facts': + return {'images': [], 'failed': available_locally} + return {'changed': False} + + result = DockerImageAvailability(execute_module=execute_module).run(tmp=None, task_vars=dict( + openshift=dict( + common=dict( service_type='origin', is_containerized=False, - )), - openshift_release='v3.5', - deployment_type='origin', - openshift_image_tag='', # FIXME: should not be required + is_atomic=False, + ), + docker=dict(additional_registries=["docker.io", "registry.access.redhat.com"]), ), - {'changed': False}, + openshift_deployment_type='origin', + openshift_release='3.4', + openshift_image_tag='v3.4', + )) + + assert not result.get('failed', False) + + +def test_all_images_unavailable(): + def execute_module(module_name=None, module_args=None, tmp=None, task_vars=None): + if module_name == "command": + return { + 'failed': True, + } + + return { + 'changed': False, + } + + check = DockerImageAvailability(execute_module=execute_module) + actual = check.run(tmp=None, task_vars=dict( + openshift=dict( + common=dict( + service_type='origin', + is_containerized=False, + is_atomic=False, + ), + docker=dict(additional_registries=["docker.io"]), + ), + openshift_deployment_type="openshift-enterprise", + openshift_release=None, + openshift_image_tag='latest' + )) + + assert actual['failed'] + assert "required Docker images are not available" in actual['msg'] + + +@pytest.mark.parametrize("message,extra_words", [ + ( + "docker image update failure", + ["docker image update failure"], + ), + ( + "No package matching 'skopeo' found available, installed or updated", + ["dependencies can be installed via `yum`"] ), - # TODO: add more parameters here to test the multiple possible inputs that affect behavior. ]) -def test_docker_image_availability(task_vars, expected_result): +def test_skopeo_update_failure(message, extra_words): def execute_module(module_name=None, module_args=None, tmp=None, task_vars=None): - return {'info': {}} # TODO: this will vary depending on input parameters. + if module_name == "yum": + return { + "failed": True, + "msg": message, + "changed": False, + } - check = DockerImageAvailability(execute_module=execute_module) - result = check.run(tmp=None, task_vars=task_vars) - assert result == expected_result + return {'changed': False} + + actual = DockerImageAvailability(execute_module=execute_module).run(tmp=None, task_vars=dict( + openshift=dict( + common=dict( + service_type='origin', + is_containerized=False, + is_atomic=False, + ), + docker=dict(additional_registries=["unknown.io"]), + ), + openshift_deployment_type="openshift-enterprise", + openshift_release='', + openshift_image_tag='', + )) + + assert actual["failed"] + for word in extra_words: + assert word in actual["msg"] + + +@pytest.mark.parametrize("deployment_type,registries", [ + ("origin", ["unknown.io"]), + ("openshift-enterprise", ["registry.access.redhat.com"]), + ("openshift-enterprise", []), +]) +def test_registry_availability(deployment_type, registries): + def execute_module(module_name=None, module_args=None, tmp=None, task_vars=None): + return { + 'changed': False, + } + + actual = DockerImageAvailability(execute_module=execute_module).run(tmp=None, task_vars=dict( + openshift=dict( + common=dict( + service_type='origin', + is_containerized=False, + is_atomic=False, + ), + docker=dict(additional_registries=registries), + ), + openshift_deployment_type=deployment_type, + openshift_release='', + openshift_image_tag='', + )) + + assert not actual.get("failed", False) diff --git a/roles/openshift_health_checker/test/memory_availability_test.py b/roles/openshift_health_checker/test/memory_availability_test.py index e161a5b9e..4fbaea0a9 100644 --- a/roles/openshift_health_checker/test/memory_availability_test.py +++ b/roles/openshift_health_checker/test/memory_availability_test.py @@ -20,27 +20,42 @@ def test_is_active(group_names, is_active): assert MemoryAvailability.is_active(task_vars=task_vars) == is_active -@pytest.mark.parametrize('group_names,ansible_memtotal_mb', [ +@pytest.mark.parametrize('group_names,configured_min,ansible_memtotal_mb', [ ( ['masters'], + 0, 17200, ), ( ['nodes'], + 0, 8200, ), ( + ['nodes'], + 1, # configure lower threshold + 2000, # too low for recommended but not for configured + ), + ( + ['nodes'], + 2, # configure threshold where adjustment pushes it over + 1900, + ), + ( ['etcd'], - 22200, + 0, + 8200, ), ( ['masters', 'nodes'], + 0, 17000, ), ]) -def test_succeeds_with_recommended_memory(group_names, ansible_memtotal_mb): +def test_succeeds_with_recommended_memory(group_names, configured_min, ansible_memtotal_mb): task_vars = dict( group_names=group_names, + openshift_check_min_host_memory_gb=configured_min, ansible_memtotal_mb=ansible_memtotal_mb, ) @@ -50,39 +65,62 @@ def test_succeeds_with_recommended_memory(group_names, ansible_memtotal_mb): assert not result.get('failed', False) -@pytest.mark.parametrize('group_names,ansible_memtotal_mb,extra_words', [ +@pytest.mark.parametrize('group_names,configured_min,ansible_memtotal_mb,extra_words', [ ( ['masters'], 0, - ['0.0 GB'], + 0, + ['0.0 GiB'], ), ( ['nodes'], + 0, 100, - ['0.1 GB'], + ['0.1 GiB'], + ), + ( + ['nodes'], + 24, # configure higher threshold + 20 * 1024, # enough to meet recommended but not configured + ['20.0 GiB'], + ), + ( + ['nodes'], + 24, # configure higher threshold + 22 * 1024, # not enough for adjustment to push over threshold + ['22.0 GiB'], ), ( ['etcd'], - -1, - ['0.0 GB'], + 0, + 6 * 1024, + ['6.0 GiB'], + ), + ( + ['etcd', 'masters'], + 0, + 9 * 1024, # enough memory for etcd, not enough for a master + ['9.0 GiB'], ), ( ['nodes', 'masters'], + 0, # enough memory for a node, not enough for a master - 11000, - ['11.0 GB'], + 11 * 1024, + ['11.0 GiB'], ), ]) -def test_fails_with_insufficient_memory(group_names, ansible_memtotal_mb, extra_words): +def test_fails_with_insufficient_memory(group_names, configured_min, ansible_memtotal_mb, extra_words): task_vars = dict( group_names=group_names, + openshift_check_min_host_memory_gb=configured_min, ansible_memtotal_mb=ansible_memtotal_mb, ) check = MemoryAvailability(execute_module=fake_execute_module) result = check.run(tmp=None, task_vars=task_vars) - assert result['failed'] + assert result.get('failed', False) for word in 'below recommended'.split() + extra_words: assert word in result['msg'] |