def test_no_sources(self, hostname, ipaddress, mode, config_cache, host_config): broker = ParsedSectionsBroker() update_host_sections( broker, make_nodes( config_cache, host_config, ipaddress, mode=mode, sources=(), ), max_cachefile_age=0, host_config=host_config, fetcher_messages=(), selected_sections=NO_SELECTION, ) # The length is not zero because the function always sets, # at least, a piggy back section. assert len(broker) == 1 key = HostKey(hostname, ipaddress, SourceType.HOST) assert key in broker section = broker[key] # Public attributes from HostSections: assert not section.sections assert not section.cache_info assert not section.piggybacked_raw_data
def test_multiple_sources_from_the_same_host( self, hostname, ipaddress, mode, config_cache, host_config, ): sources = [ ProgramSource.ds( hostname, ipaddress, mode=mode, template="", ), TCPSource( hostname, ipaddress, mode=mode, ), ] broker = ParsedSectionsBroker() update_host_sections( broker, make_nodes( config_cache, host_config, ipaddress, mode=mode, sources=sources, ), max_cachefile_age=0, host_config=host_config, fetcher_messages=[ FetcherMessage.from_raw_data( result.OK(source.default_raw_data), Snapshot.null(), source.fetcher_type, ) for source in sources ], selected_sections=NO_SELECTION, ) assert len(broker) == 1 key = HostKey(hostname, ipaddress, SourceType.HOST) assert key in broker section = broker[key] assert len(section.sections) == 1 # yapf: disable assert (section.sections[SectionName("section_name_%s" % hostname)] == len(sources) * [["section_content"]])
def test_one_snmp_source(self, hostname, ipaddress, mode, config_cache, host_config): broker = ParsedSectionsBroker() update_host_sections( broker, make_nodes( config_cache, host_config, ipaddress, mode=mode, sources=[ SNMPSource.snmp( hostname, ipaddress, mode=mode, selected_sections=NO_SELECTION, on_scan_error="raise", ), ], ), max_cachefile_age=0, host_config=host_config, fetcher_messages=[ FetcherMessage.from_raw_data( result.OK({}), Snapshot.null(), FetcherType.SNMP, ), ], selected_sections=NO_SELECTION, ) assert len(broker) == 1 key = HostKey(hostname, ipaddress, SourceType.HOST) assert key in broker section = broker[key] assert len(section.sections) == 1 assert section.sections[SectionName("section_name_%s" % hostname)] == [["section_content"]]
def test_no_sources(self, cluster, nodes, config_cache, host_config, mode): made_nodes = make_nodes( config_cache, host_config, None, mode=mode, sources=(), ) broker = ParsedSectionsBroker() update_host_sections( broker, made_nodes, max_cachefile_age=0, host_config=host_config, fetcher_messages=[ # We do not pass sources explicitly but still append Piggyback. FetcherMessage.from_raw_data( result.OK(AgentRawData(b"")), Snapshot.null(), FetcherType.PIGGYBACK, ) for _n in made_nodes ], selected_sections=NO_SELECTION, ) assert len(broker) == len(nodes) key_clu = HostKey(cluster, None, SourceType.HOST) assert key_clu not in broker for hostname, addr in nodes.items(): key = HostKey(hostname, addr, SourceType.HOST) assert key in broker section = broker[key] # yapf: disable assert (section.sections[SectionName("section_name_%s" % hostname)] == [["section_content_%s" % hostname]]) assert not section.cache_info assert not section.piggybacked_raw_data
def test_one_nonsnmp_source(self, hostname, ipaddress, mode, config_cache, host_config, source): source = source(hostname, ipaddress, mode=mode) assert source.source_type is SourceType.HOST broker = ParsedSectionsBroker() update_host_sections( broker, make_nodes( config_cache, host_config, ipaddress, mode=mode, sources=[source], ), max_cachefile_age=0, host_config=host_config, fetcher_messages=[ FetcherMessage.from_raw_data( result.OK(source.default_raw_data), Snapshot.null(), source.fetcher_type, ), ], selected_sections=NO_SELECTION, ) assert len(broker) == 1 key = HostKey(hostname, ipaddress, source.source_type) assert key in broker section = broker[key] assert len(section.sections) == 1 assert section.sections[SectionName("section_name_%s" % hostname)] == [["section_content"]]
def _fetch_parsed_sections_broker_for_inv( config_cache: config.ConfigCache, host_config: config.HostConfig, ipaddress: Optional[HostAddress], selected_sections: SectionNameCollection, ) -> Tuple[ParsedSectionsBroker, Sequence[Tuple[Source, result.Result[ HostSections, Exception]]]]: if host_config.is_cluster: return ParsedSectionsBroker(), [] mode = (Mode.INVENTORY if selected_sections is NO_SELECTION else Mode.FORCE_SECTIONS) nodes = sources.make_nodes( config_cache, host_config, ipaddress, mode, sources.make_sources( host_config, ipaddress, mode=mode, selected_sections=selected_sections, ), ) parsed_sections_broker = ParsedSectionsBroker() results = sources.update_host_sections( parsed_sections_broker, nodes, max_cachefile_age=host_config.max_cachefile_age, host_config=host_config, fetcher_messages=list( sources.fetch_all( nodes, max_cachefile_age=host_config.max_cachefile_age, host_config=host_config, )), selected_sections=selected_sections, ) return parsed_sections_broker, results
def test_get_host_sections_cluster(mode, monkeypatch, mocker): hostname = "testhost" hosts = { "host0": "10.0.0.0", "host1": "10.0.0.1", "host2": "10.0.0.2", } address = "1.2.3.4" tags = {"agent": "no-agent"} section_name = SectionName("test_section") config_cache = make_scenario(hostname, tags).apply(monkeypatch) host_config = config.HostConfig.make_host_config(hostname) def lookup_ip_address(host_config, family=None, for_mgmt_board=False): return hosts[host_config.hostname] def check(_, *args, **kwargs): return result.OK(AgentHostSections(sections={section_name: [[str(section_name)]]})) monkeypatch.setattr( ip_lookup, "lookup_ip_address", lookup_ip_address, ) monkeypatch.setattr( Source, "parse", check, ) mocker.patch.object( cmk.utils.piggyback, "remove_source_status_file", autospec=True, ) mocker.patch.object( cmk.utils.piggyback, "_store_status_file_of", autospec=True, ) # Create a cluster host_config.nodes = list(hosts.keys()) nodes = make_nodes( config_cache, host_config, address, mode=mode, sources=make_sources(host_config, address, mode=mode) ) broker = ParsedSectionsBroker() update_host_sections( broker, nodes, max_cachefile_age=host_config.max_cachefile_age, host_config=host_config, fetcher_messages=[ FetcherMessage.from_raw_data( result.OK(source.default_raw_data), Snapshot.null(), source.fetcher_type, ) for _h, _i, sources in nodes for source in sources ], selected_sections=NO_SELECTION, ) assert len(broker) == len(hosts) == 3 cmk.utils.piggyback._store_status_file_of.assert_not_called() # type: ignore[attr-defined] assert cmk.utils.piggyback.remove_source_status_file.call_count == 3 # type: ignore[attr-defined] for host, addr in hosts.items(): remove_source_status_file = cmk.utils.piggyback.remove_source_status_file remove_source_status_file.assert_any_call(host) # type: ignore[attr-defined] key = HostKey(host, addr, SourceType.HOST) assert key in broker section = broker[key] assert len(section.sections) == 1 assert next(iter(section.sections)) == section_name assert not section.cache_info assert not section.piggybacked_raw_data
def do_check( hostname: HostName, ipaddress: Optional[HostAddress], *, # The following arguments *must* remain optional for Nagios and the `DiscoCheckExecutor`. # See Also: `cmk.base.discovery.check_discovery()` fetcher_messages: Sequence[FetcherMessage] = (), run_only_plugin_names: Optional[Set[CheckPluginName]] = None, selected_sections: SectionNameCollection = NO_SELECTION, submit_to_core: bool = True, show_perfdata: bool = False, ) -> Tuple[int, List[ServiceDetails], List[ServiceAdditionalDetails], List[str]]: console.verbose("Checkmk version %s\n", cmk_version.__version__) config_cache = config.get_config_cache() host_config = config_cache.get_host_config(hostname) exit_spec = host_config.exit_code_spec() mode = Mode.CHECKING if selected_sections is NO_SELECTION else Mode.FORCE_SECTIONS status: ServiceState = 0 infotexts: List[ServiceDetails] = [] long_infotexts: List[ServiceAdditionalDetails] = [] perfdata: List[str] = [] try: license_usage.try_history_update() # In case of keepalive we always have an ipaddress (can be 0.0.0.0 or :: when # address is unknown). When called as non keepalive ipaddress may be None or # is already an address (2nd argument) if ipaddress is None and not host_config.is_cluster: ipaddress = ip_lookup.lookup_ip_address(host_config) # When monitoring Checkmk clusters, the cluster nodes are responsible for fetching all # information from the monitored host and cache the result for the cluster checks to be # performed on the cached information. # # This means that in case of SNMP nodes, they need to take the clustered services of the # node into account, fetch the needed sections and cache them for the cluster host. # # But later, when checking the node services, the node has to only deal with the unclustered # services. # # TODO: clean this up. The fetched sections are computed in the checkers # _make_configured_snmp_sections now. # belongs_to_cluster = len(config_cache.clusters_of(hostname)) > 0 services_to_fetch = _get_services_to_fetch( host_name=hostname, belongs_to_cluster=belongs_to_cluster, config_cache=config_cache, ) services_to_check = _filter_clustered_services( config_cache=config_cache, host_name=hostname, belongs_to_cluster=belongs_to_cluster, services=services_to_fetch, run_only_plugin_names=run_only_plugin_names, ) nodes = sources.make_nodes( config_cache, host_config, ipaddress, mode, sources.make_sources( host_config, ipaddress, mode=mode, selected_sections=selected_sections, ), ) if not fetcher_messages: # Note: `fetch_all(sources)` is almost always called in similar # code in discovery and inventory. The only other exception # is `cmk.base.discovery.check_discovery(...)`. This does # not seem right. fetcher_messages = list( sources.fetch_all( nodes, max_cachefile_age=host_config.max_cachefile_age, host_config=host_config, )) with CPUTracker() as tracker: broker = ParsedSectionsBroker() result = sources.update_host_sections( broker, nodes, max_cachefile_age=host_config.max_cachefile_age, host_config=host_config, fetcher_messages=fetcher_messages, selected_sections=selected_sections, ) num_success, plugins_missing_data = _do_all_checks_on_host( config_cache, host_config, ipaddress, parsed_sections_broker=broker, services=services_to_check, submit_to_core=submit_to_core, show_perfdata=show_perfdata, ) if run_only_plugin_names is None: inventory.do_inventory_actions_during_checking_for( config_cache, host_config, ipaddress, parsed_sections_broker=broker, ) for source, host_sections in result: source_state, source_output, source_perfdata = source.summarize( host_sections) if source_output != "": status = max(status, source_state) infotexts.append("[%s] %s" % (source.id, source_output)) perfdata.extend( [_convert_perf_data(p) for p in source_perfdata]) if plugins_missing_data: missing_data_status, missing_data_infotext = _check_plugins_missing_data( plugins_missing_data, exit_spec, bool(num_success), ) status = max(status, missing_data_status) infotexts.append(missing_data_infotext) total_times = tracker.duration for msg in fetcher_messages: total_times += msg.stats.duration infotexts.append("execution time %.1f sec" % total_times.process.elapsed) if config.check_mk_perfdata_with_times: perfdata += [ "execution_time=%.3f" % total_times.process.elapsed, "user_time=%.3f" % total_times.process.user, "system_time=%.3f" % total_times.process.system, "children_user_time=%.3f" % total_times.process.children_user, "children_system_time=%.3f" % total_times.process.children_system, ] summary: DefaultDict[str, Snapshot] = defaultdict(Snapshot.null) for msg in fetcher_messages if fetcher_messages else (): if msg.fetcher_type in ( FetcherType.PIGGYBACK, FetcherType.PROGRAM, FetcherType.SNMP, FetcherType.TCP, ): summary[{ FetcherType.PIGGYBACK: "agent", FetcherType.PROGRAM: "ds", FetcherType.SNMP: "snmp", FetcherType.TCP: "agent", }[msg.fetcher_type]] += msg.stats.duration for phase, duration in summary.items(): perfdata.append("cmk_time_%s=%.3f" % (phase, duration.idle)) else: perfdata.append("execution_time=%.3f" % total_times.process.elapsed) return status, infotexts, long_infotexts, perfdata finally: if _checkresult_file_fd is not None: _close_checkresult_file()
def check_discovery( host_name: HostName, ipaddress: Optional[HostAddress], *, # The next argument *must* remain optional for the DiscoCheckExecutor. # See Also: `cmk.base.checking.do_check()`. fetcher_messages: Sequence[FetcherMessage] = (), ) -> Tuple[int, List[str], List[str], List[Tuple]]: # Note: '--cache' is set in core_cmc, nagios template or even on CL and means: # 1. use caches as default: # - Set FileCacheFactory.maybe = True (set max_cachefile_age, else 0) # - Set FileCacheFactory.use_outdated = True # 2. Then these settings are used to read cache file or not config_cache = config.get_config_cache() host_config = config_cache.get_host_config(host_name) discovery_parameters = DiscoveryParameters( on_error="raise", load_labels=True, save_labels=False, only_host_labels=False, ) params = host_config.discovery_check_parameters if params is None: params = host_config.default_discovery_check_parameters() # In case of keepalive discovery we always have an ipaddress. When called as non keepalive # ipaddress is always None if ipaddress is None and not host_config.is_cluster: ipaddress = config.lookup_ip_address(host_config) nodes = sources.make_nodes( config_cache, host_config, ipaddress, Mode.DISCOVERY, sources.make_sources(host_config, ipaddress, mode=Mode.DISCOVERY), ) use_caches = cmk.core_helpers.cache.FileCacheFactory.maybe max_cachefile_age = config.discovery_max_cachefile_age() if use_caches else 0 if not fetcher_messages: # Note: *Not* calling `fetch_all(sources)` here is probably buggy. # Also See: `cmk.base.checking.do_check()` fetcher_messages = list( sources.fetch_all( nodes, max_cachefile_age=max_cachefile_age, host_config=host_config, )) parsed_sections_broker = ParsedSectionsBroker() result = sources.update_host_sections( parsed_sections_broker, nodes, max_cachefile_age=max_cachefile_age, host_config=host_config, fetcher_messages=fetcher_messages, selected_sections=NO_SELECTION, ) services, host_label_discovery_result = _get_host_services( host_config, ipaddress, parsed_sections_broker, discovery_parameters, ) status, infotexts, long_infotexts, perfdata, need_rediscovery = _aggregate_subresults( _check_service_lists(host_name, services, params), _check_host_labels(host_label_discovery_result, params), _check_data_sources(result), ) if need_rediscovery: if host_config.is_cluster and host_config.nodes: for nodename in host_config.nodes: _set_rediscovery_flag(nodename) else: _set_rediscovery_flag(host_name) infotexts.append(u"rediscovery scheduled") return status, infotexts, long_infotexts, perfdata
def discover_on_host( *, config_cache: config.ConfigCache, host_config: config.HostConfig, mode: str, service_filters: Optional[_ServiceFilters], on_error: str, use_cached_snmp_data: bool, max_cachefile_age: int, ) -> DiscoveryResult: console.verbose(" Doing discovery with mode '%s'...\n" % mode) host_name = host_config.hostname result = DiscoveryResult() discovery_parameters = DiscoveryParameters( on_error=on_error, load_labels=(mode != "remove"), save_labels=(mode != "remove"), only_host_labels=(mode == "only-host-labels"), ) if host_name not in config_cache.all_active_hosts(): result.error_text = "" return result _set_cache_opts_of_checkers(use_cached_snmp_data=use_cached_snmp_data) try: # in "refresh" mode we first need to remove all previously discovered # checks of the host, so that _get_host_services() does show us the # new discovered check parameters. if mode == "refresh": result.self_removed += host_config.remove_autochecks() # this is cluster-aware! if host_config.is_cluster: ipaddress = None else: ipaddress = config.lookup_ip_address(host_config) nodes = sources.make_nodes( config_cache, host_config, ipaddress, Mode.DISCOVERY, sources.make_sources( host_config, ipaddress, mode=Mode.DISCOVERY, on_scan_error=on_error, ), ) parsed_sections_broker = ParsedSectionsBroker() sources.update_host_sections( parsed_sections_broker, nodes, max_cachefile_age=max_cachefile_age, host_config=host_config, fetcher_messages=list( sources.fetch_all( nodes, max_cachefile_age=max_cachefile_age, host_config=host_config, )), selected_sections=NO_SELECTION, ) # Compute current state of new and existing checks services, host_labels = _get_host_services( host_config, ipaddress, parsed_sections_broker, discovery_parameters, ) old_services = services.get("old", []) # Create new list of checks new_services = _get_post_discovery_services(host_name, services, service_filters or _ServiceFilters.accept_all(), result, mode) host_config.set_autochecks(new_services) result.diff_text = make_object_diff( _make_services_audit_log_object([x.service for x in old_services]), _make_services_audit_log_object([x.service for x in new_services])) except MKTimeout: raise # let general timeout through except Exception as e: if cmk.utils.debug.enabled(): raise result.error_text = str(e) else: if mode != "remove": result.self_new_host_labels = len(host_labels.new) result.self_total_host_labels = len(host_labels.present) result.self_total = result.self_new + result.self_kept return result
def do_discovery( arg_hostnames: Set[HostName], *, selected_sections: SectionNameCollection, run_only_plugin_names: Optional[Set[CheckPluginName]], arg_only_new: bool, only_host_labels: bool = False, ) -> None: config_cache = config.get_config_cache() use_caches = not arg_hostnames or cmk.core_helpers.cache.FileCacheFactory.maybe on_error = "raise" if cmk.utils.debug.enabled() else "warn" discovery_parameters = DiscoveryParameters( on_error=on_error, load_labels=arg_only_new, save_labels=True, only_host_labels=only_host_labels, ) host_names = _preprocess_hostnames(arg_hostnames, config_cache, only_host_labels) mode = Mode.DISCOVERY if selected_sections is NO_SELECTION else Mode.FORCE_SECTIONS # Now loop through all hosts for host_name in sorted(host_names): host_config = config_cache.get_host_config(host_name) section.section_begin(host_name) try: ipaddress = config.lookup_ip_address(host_config) nodes = sources.make_nodes( config_cache, host_config, ipaddress, mode, sources.make_sources( host_config, ipaddress, mode=mode, selected_sections=selected_sections, on_scan_error=on_error, ), ) max_cachefile_age = config.discovery_max_cachefile_age() if use_caches else 0 parsed_sections_broker = ParsedSectionsBroker() sources.update_host_sections( parsed_sections_broker, nodes, max_cachefile_age=max_cachefile_age, host_config=host_config, fetcher_messages=list( sources.fetch_all( nodes, max_cachefile_age=max_cachefile_age, host_config=host_config, )), selected_sections=selected_sections, ) _do_discovery_for( host_name, ipaddress, parsed_sections_broker, run_only_plugin_names, arg_only_new, discovery_parameters, ) except Exception as e: if cmk.utils.debug.enabled(): raise section.section_error("%s" % e) finally: cmk.utils.cleanup.cleanup_globals()
def get_check_preview( *, host_name: HostName, max_cachefile_age: int, use_cached_snmp_data: bool, on_error: str, ) -> Tuple[CheckPreviewTable, DiscoveredHostLabels]: """Get the list of service of a host or cluster and guess the current state of all services if possible""" config_cache = config.get_config_cache() host_config = config_cache.get_host_config(host_name) ip_address = None if host_config.is_cluster else config.lookup_ip_address(host_config) discovery_parameters = DiscoveryParameters( on_error=on_error, load_labels=False, save_labels=False, only_host_labels=False, ) _set_cache_opts_of_checkers(use_cached_snmp_data=use_cached_snmp_data) nodes = sources.make_nodes( config_cache, host_config, ip_address, Mode.DISCOVERY, sources.make_sources( host_config, ip_address, mode=Mode.DISCOVERY, on_scan_error=on_error, )) parsed_sections_broker = ParsedSectionsBroker() sources.update_host_sections( parsed_sections_broker, nodes, max_cachefile_age=max_cachefile_age, host_config=host_config, fetcher_messages=list( sources.fetch_all( nodes, max_cachefile_age=max_cachefile_age, host_config=host_config, )), selected_sections=NO_SELECTION, ) grouped_services, host_label_result = _get_host_services( host_config, ip_address, parsed_sections_broker, discovery_parameters, ) table: CheckPreviewTable = [] for check_source, services_with_nodes in grouped_services.items(): for service, found_on_nodes in services_with_nodes: plugin = agent_based_register.get_check_plugin(service.check_plugin_name) params = _preview_params(host_name, service, plugin, check_source) if check_source in ['legacy', 'active', 'custom']: exitcode = None output = u"WAITING - %s check, cannot be done offline" % check_source.title() ruleset_name: Optional[RulesetName] = None else: ruleset_name = (str(plugin.check_ruleset_name) if plugin and plugin.check_ruleset_name else None) wrapped_params = (Parameters(wrap_parameters(params)) if plugin and plugin.check_default_parameters is not None else None) exitcode, output, _perfdata = checking.get_aggregated_result( parsed_sections_broker, host_config, ip_address, service, plugin, lambda p=wrapped_params: p, # type: ignore[misc] # "type of lambda" ).result # Service discovery never uses the perfdata in the check table. That entry # is constantly discarded, yet passed around(back and forth) as part of the # discovery result in the request elements. Some perfdata VALUES are not parsable # by ast.literal_eval such as "inf" it lead to ValueErrors. Thus keep perfdata empty perfdata: List[MetricTuple] = [] table.append(( _preview_check_source(host_name, service, check_source), str(service.check_plugin_name), ruleset_name, service.item, service.parameters, params, service.description, exitcode, output, perfdata, service.service_labels.to_dict(), found_on_nodes, )) return table, DiscoveredHostLabels( *{ # TODO (mo): According to unit tests, this is what was done prior to refactoring. # Im not sure this is desired. If it is, it should be explained. **{l.name: l for l in host_label_result.vanished}, **{l.name: l for l in host_label_result.present}, }.values())