def do_check( hostname: HostName, ipaddress: Optional[HostAddress], only_check_plugin_names: Optional[Set[CheckPluginName]] = None, fetcher_messages: Optional[Sequence[FetcherMessage]] = None ) -> 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() status: ServiceState = 0 infotexts: List[ServiceDetails] = [] long_infotexts: List[ServiceAdditionalDetails] = [] perfdata: List[str] = [] try: with cpu_tracking.execute(), cpu_tracking.phase("busy"): 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) item_state.load(hostname) # 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. 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, only_check_plugins=only_check_plugin_names, ) services_to_check = _filter_clustered_services( config_cache=config_cache, host_name=hostname, belongs_to_cluster=belongs_to_cluster, services=services_to_fetch, ) # see which raw sections we may need selected_raw_sections = agent_based_register.get_relevant_raw_sections( check_plugin_names=(s.check_plugin_name for s in services_to_fetch), consider_inventory_plugins=host_config.do_status_data_inventory, ) sources = checkers.make_sources( host_config, ipaddress, mode=checkers.Mode.CHECKING, ) mhs = MultiHostSections() result = checkers.update_host_sections( mhs, checkers.make_nodes( config_cache, host_config, ipaddress, checkers.Mode.CHECKING, sources, ), selected_raw_sections=selected_raw_sections, max_cachefile_age=host_config.max_cachefile_age, host_config=host_config, fetcher_messages=fetcher_messages, ) num_success, plugins_missing_data = _do_all_checks_on_host( config_cache, host_config, ipaddress, multi_host_sections=mhs, services=services_to_check, only_check_plugins=only_check_plugin_names, ) inventory.do_inventory_actions_during_checking_for( config_cache, host_config, ipaddress, sources=sources, multi_host_sections=mhs, ) if _submit_to_core: item_state.save(hostname) 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) for msg in fetcher_messages if fetcher_messages else (): cpu_tracking.update(msg.stats.cpu_times) phase_times = cpu_tracking.get_times() total_times = phase_times["busy"] infotexts.append("execution time %.1f sec" % total_times.run_time) if config.check_mk_perfdata_with_times: perfdata += [ "execution_time=%.3f" % total_times.run_time, "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, ] for phase, times in phase_times.items(): if phase in ["agent", "snmp", "ds"]: t = times.run_time - sum(times.process[:4]) # real time - CPU time perfdata.append("cmk_time_%s=%.3f" % (phase, t)) else: perfdata.append("execution_time=%.3f" % total_times.run_time) return status, infotexts, long_infotexts, perfdata finally: if _checkresult_file_fd is not None: _close_checkresult_file() # "ipaddress is not None": At least when working with a cluster host it seems the ipaddress # may be None. This needs to be understood in detail and cleaned up. As the InlineSNMP # stats feature is a very rarely used debugging feature, the analyzation and fix is # postponed now. if config.record_inline_snmp_stats and ipaddress is not None and host_config.snmp_config( ipaddress).snmp_backend == "inline": inline.snmp_stats_save()
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: checkers.SectionNameCollection = checkers.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 = checkers.Mode.CHECKING if selected_sections is checkers.NO_SELECTION else checkers.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 = checkers.make_nodes( config_cache, host_config, ipaddress, mode, checkers.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( checkers.fetch_all( nodes, max_cachefile_age=host_config.max_cachefile_age, host_config=host_config, )) with CPUTracker() as tracker: mhs = MultiHostSections() result = checkers.update_host_sections( mhs, 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, multi_host_sections=mhs, 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, multi_host_sections=mhs, ) 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 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_plugin_names: Container[CheckPluginName] = EVERYTHING, selected_sections: SectionNameCollection = NO_SELECTION, dry_run: bool = False, show_perfdata: bool = False, ) -> Tuple[int, List[ServiceDetails], List[ServiceAdditionalDetails], List[str]]: console.vverbose("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 = config.lookup_ip_address(host_config) services_to_check = _get_services_to_check( config_cache=config_cache, host_name=hostname, run_plugin_names=run_plugin_names, ) with CPUTracker() as tracker: broker, source_results = make_broker( config_cache=config_cache, host_config=host_config, ip_address=ipaddress, mode=mode, selected_sections=selected_sections, file_cache_max_age=host_config.max_cachefile_age, fetcher_messages=fetcher_messages, force_snmp_cache_refresh=False, on_scan_error="raise", ) num_success, plugins_missing_data = _do_all_checks_on_host( config_cache, host_config, ipaddress, parsed_sections_broker=broker, services=services_to_check, dry_run=dry_run, show_perfdata=show_perfdata, ) if run_plugin_names is EVERYTHING: inventory.do_inventory_actions_during_checking_for( config_cache, host_config, ipaddress, parsed_sections_broker=broker, ) for source, host_sections in source_results: source_state, source_output = source.summarize(host_sections) if source_output != "": status = worst_service_state(status, source_state, default=3) infotexts.append("[%s] %s" % (source.id, source_output)) 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: _submit_to_core.finalize()
def _execute_checkmk_checks( *, hostname: HostName, ipaddress: Optional[HostAddress], fetcher_messages: Sequence[FetcherMessage] = (), run_plugin_names: Container[CheckPluginName], selected_sections: SectionNameCollection, dry_run: bool, show_perfdata: bool, ) -> ActiveCheckResult: 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 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 = config.lookup_ip_address(host_config) services = config.resolve_service_dependencies( host_name=hostname, services=sorted( check_table.get_check_table(hostname).values(), key=lambda service: service.description, ), ) with CPUTracker() as tracker: broker, source_results = make_broker( config_cache=config_cache, host_config=host_config, ip_address=ipaddress, mode=mode, selected_sections=selected_sections, file_cache_max_age=host_config.max_cachefile_age, fetcher_messages=fetcher_messages, force_snmp_cache_refresh=False, on_scan_error=OnError.RAISE, ) num_success, plugins_missing_data = check_host_services( config_cache=config_cache, host_config=host_config, ipaddress=ipaddress, parsed_sections_broker=broker, services=services, run_plugin_names=run_plugin_names, dry_run=dry_run, show_perfdata=show_perfdata, ) if run_plugin_names is EVERYTHING: inventory.do_inventory_actions_during_checking_for( config_cache, host_config, ipaddress, parsed_sections_broker=broker, ) timed_results = [ *check_sources( source_results=source_results, mode=mode, include_ok_results=True, ), *check_parsing_errors(errors=broker.parsing_errors(), ), *_check_plugins_missing_data( plugins_missing_data, exit_spec, bool(num_success), ), ] return ActiveCheckResult.from_subresults( *timed_results, _timing_results(tracker, fetcher_messages), ) finally: _submit_to_core.finalize()
def _execute_checkmk_checks( *, hostname: HostName, ipaddress: Optional[HostAddress], fetched: Sequence[Tuple[Source, FetcherMessage]], run_plugin_names: Container[CheckPluginName], selected_sections: SectionNameCollection, dry_run: bool, show_perfdata: bool, ) -> ActiveCheckResult: config_cache = config.get_config_cache() host_config = config_cache.get_host_config(hostname) exit_spec = host_config.exit_code_spec() try: license_usage.try_history_update() services = config.resolve_service_dependencies( host_name=hostname, services=sorted( check_table.get_check_table(hostname).values(), key=lambda service: service.description, ), ) broker, source_results = make_broker( fetched=fetched, selected_sections=selected_sections, file_cache_max_age=host_config.max_cachefile_age, ) with CPUTracker() as tracker: num_success, plugins_missing_data = check_host_services( config_cache=config_cache, host_config=host_config, ipaddress=ipaddress, parsed_sections_broker=broker, services=services, run_plugin_names=run_plugin_names, dry_run=dry_run, show_perfdata=show_perfdata, ) if run_plugin_names is EVERYTHING: inventory.do_inventory_actions_during_checking_for( config_cache, host_config, parsed_sections_broker=broker, ) timed_results = [ *check_sources( source_results=source_results, include_ok_results=True, ), *check_parsing_errors(errors=broker.parsing_errors(), ), *_check_plugins_missing_data( plugins_missing_data, exit_spec, bool(num_success), ), ] return ActiveCheckResult.from_subresults( *timed_results, _timing_results(tracker.duration, [fetched_entry[1] for fetched_entry in fetched]), ) finally: _submit_to_core.finalize()