diff options
Diffstat (limited to 'roles/openshift_health_checker/action_plugins/openshift_health_check.py')
| -rw-r--r-- | roles/openshift_health_checker/action_plugins/openshift_health_check.py | 146 |
1 files changed, 106 insertions, 40 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 05e53333d..8d35db6b5 100644 --- a/roles/openshift_health_checker/action_plugins/openshift_health_check.py +++ b/roles/openshift_health_checker/action_plugins/openshift_health_check.py @@ -1,76 +1,74 @@ """ Ansible action plugin to execute health checks in OpenShift clusters. """ -# pylint: disable=wrong-import-position,missing-docstring,invalid-name import sys import os +import traceback from collections import defaultdict +from ansible.plugins.action import ActionBase +from ansible.module_utils.six import string_types + try: from __main__ import display except ImportError: + # pylint: disable=ungrouped-imports; this is the standard way how to import + # the default display object in Ansible action plugins. from ansible.utils.display import Display display = Display() -from ansible.plugins.action import ActionBase -from ansible.module_utils.six import string_types - # Augment sys.path so that we can import checks from a directory relative to # this callback plugin. sys.path.insert(1, os.path.dirname(os.path.dirname(__file__))) +# pylint: disable=wrong-import-position; the import statement must come after +# the manipulation of sys.path. from openshift_checks import OpenShiftCheck, OpenShiftCheckException, load_checks # noqa: E402 class ActionModule(ActionBase): + """Action plugin to execute health checks.""" def run(self, tmp=None, task_vars=None): result = super(ActionModule, self).run(tmp, task_vars) task_vars = task_vars or {} - # vars are not supportably available in the callback plugin, - # so record any it will need in the result. + # callback plugins cannot read Ansible vars, but we would like + # zz_failure_summary to have access to certain values. We do so by + # storing the information we 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 - result["msg"] = "'openshift' is undefined, did 'openshift_facts' run?" - return result - try: known_checks = self.load_known_checks(tmp, task_vars) args = self._task.args requested_checks = normalize(args.get('checks', [])) + + if not requested_checks: + result['failed'] = True + result['msg'] = list_known_checks(known_checks) + return result + resolved_checks = resolve_checks(requested_checks, known_checks.values()) - except OpenShiftCheckException as e: + except OpenShiftCheckException as exc: result["failed"] = True - result["msg"] = str(e) + result["msg"] = str(exc) + return result + + if "openshift" not in task_vars: + result["failed"] = True + result["msg"] = "'openshift' is undefined, did 'openshift_facts' run?" return result result["checks"] = check_results = {} user_disabled_checks = normalize(task_vars.get('openshift_disable_check', [])) - for check_name in resolved_checks: - display.banner("CHECK [{} : {}]".format(check_name, task_vars["ansible_host"])) - check = known_checks[check_name] - - if not check.is_active(): - 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() - except OpenShiftCheckException as e: - r = dict( - failed=True, - msg=str(e), - ) - + for name in resolved_checks: + display.banner("CHECK [{} : {}]".format(name, task_vars["ansible_host"])) + check = known_checks[name] + check_results[name] = run_check(name, check, user_disabled_checks) if check.changed: - r["changed"] = True - check_results[check_name] = r + check_results[name]["changed"] = True result["changed"] = any(r.get("changed") for r in check_results.values()) if any(r.get("failed") for r in check_results.values()): @@ -80,22 +78,55 @@ class ActionModule(ActionBase): return result def load_known_checks(self, tmp, task_vars): + """Find all existing checks and return a mapping of names to instances.""" load_checks() known_checks = {} for cls in OpenShiftCheck.subclasses(): - check_name = cls.name - if check_name in known_checks: - other_cls = known_checks[check_name].__class__ + name = cls.name + if name in known_checks: + other_cls = known_checks[name].__class__ raise OpenShiftCheckException( - "non-unique check name '{}' in: '{}.{}' and '{}.{}'".format( - check_name, - cls.__module__, cls.__name__, - other_cls.__module__, other_cls.__name__)) - known_checks[check_name] = cls(execute_module=self._execute_module, tmp=tmp, task_vars=task_vars) + "duplicate check name '{}' in: '{}' and '{}'" + "".format(name, full_class_name(cls), full_class_name(other_cls)) + ) + known_checks[name] = cls(execute_module=self._execute_module, tmp=tmp, task_vars=task_vars) return known_checks +def list_known_checks(known_checks): + """Return text listing the existing checks and tags.""" + # TODO: we could include a description of each check by taking it from a + # check class attribute (e.g., __doc__) when building the message below. + msg = ( + 'This playbook is meant to run health checks, but no checks were ' + 'requested. Set the `openshift_checks` variable to a comma-separated ' + 'list of check names or a YAML list. Available checks:\n {}' + ).format('\n '.join(sorted(known_checks))) + + tags = describe_tags(known_checks.values()) + + msg += ( + '\n\nTags can be used as a shortcut to select multiple ' + 'checks. Available tags and the checks they select:\n {}' + ).format('\n '.join(tags)) + + return msg + + +def describe_tags(check_classes): + """Return a sorted list of strings describing tags and the checks they include.""" + tag_checks = defaultdict(list) + for cls in check_classes: + for tag in cls.tags: + tag_checks[tag].append(cls.name) + tags = [ + '@{} = {}'.format(tag, ','.join(sorted(checks))) + for tag, checks in tag_checks.items() + ] + return sorted(tags) + + def resolve_checks(names, all_checks): """Returns a set of resolved check names. @@ -123,6 +154,12 @@ def resolve_checks(names, all_checks): if unknown_tag_names: msg.append('Unknown tag names: {}.'.format(', '.join(sorted(unknown_tag_names)))) msg.append('Make sure there is no typo in the playbook and no files are missing.') + # TODO: implement a "Did you mean ...?" when the input is similar to a + # valid check or tag. + msg.append('Known checks:') + msg.append(' {}'.format('\n '.join(sorted(known_check_names)))) + msg.append('Known tags:') + msg.append(' {}'.format('\n '.join(describe_tags(all_checks)))) raise OpenShiftCheckException('\n'.join(msg)) tag_to_checks = defaultdict(set) @@ -146,3 +183,32 @@ def normalize(checks): if isinstance(checks, string_types): checks = checks.split(',') return [name.strip() for name in checks if name.strip()] + + +def run_check(name, check, user_disabled_checks): + """Run a single check if enabled and return a result dict.""" + if name in user_disabled_checks: + return dict(skipped=True, skipped_reason="Disabled by user request") + + # pylint: disable=broad-except; capturing exceptions broadly is intentional, + # to isolate arbitrary failures in one check from others. + try: + is_active = check.is_active() + except Exception as exc: + reason = "Could not determine if check should be run, exception: {}".format(exc) + return dict(skipped=True, skipped_reason=reason, exception=traceback.format_exc()) + + if not is_active: + return dict(skipped=True, skipped_reason="Not active for this host") + + try: + return check.run() + except OpenShiftCheckException as exc: + return dict(failed=True, msg=str(exc)) + except Exception as exc: + return dict(failed=True, msg=str(exc), exception=traceback.format_exc()) + + +def full_class_name(cls): + """Return the name of a class prefixed with its module name.""" + return '{}.{}'.format(cls.__module__, cls.__name__) |