Module: check_mk Branch: master Commit: 30bcc57fce9c6fec48ece4cc0fc8e5526c532f3e URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=30bcc57fce9c6f… Author: Andreas Boesl &lt;ab(a)mathias-kettner.de&gt; Date: Mon Jan 26 16:56:57 2015 +0100 #1675 esx_vsphere_hostsystem.cpu_util_cluster: Averaged CPU utilization of all cluster nodes This new check allows you to monitor the average CPU utilization over all cluster nodes. Each node has informations about the total available MHz and the actually used MHz for computing. The actual usage of a cluster is calculated by the formula <i>(total_used_mhz / total_avail_mhz) * 100</i> Via WATO you can configure a set of WARN and CRIT levels for specific node counts. --- .werks/1675 | 13 ++++ ChangeLog | 1 + checkman/esx_vsphere_hostsystem.cpu_util_cluster | 32 +++++++++ checkman/esx_vsphere_hostsystem.mem_usage_cluster | 5 -- checks/esx_vsphere_hostsystem | 76 +++++++++++++++++++++ web/plugins/wato/check_parameters.py | 29 ++++++++ 6 files changed, 151 insertions(+), 5 deletions(-) diff --git a/.werks/1675 b/.werks/1675 new file mode 100644 index 0000000..07c644a --- /dev/null +++ b/.werks/1675 @@ -0,0 +1,13 @@ +Title: esx_vsphere_hostsystem.cpu_util_cluster: Averaged CPU utilization of all cluster nodes +Level: 1 +Component: checks +Compatible: compat +Version: 1.2.7i1 +Date: 1422287199 +Class: feature + +This new check allows you to monitor the average CPU utilization over all cluster nodes. +Each node has informations about the total available MHz and the actually used MHz for computing. +The actual usage of a cluster is calculated by the formula <i>(total_used_mhz / total_avail_mhz) * 100</i> + +Via WATO you can configure a set of WARN and CRIT levels for specific node counts. diff --git a/ChangeLog b/ChangeLog index ed3e6d1..de4d9a5 100644 --- a/ChangeLog +++ b/ChangeLog @@ -54,6 +54,7 @@ * 1857 ibm_svc_portsas: new check and extended special agent for IBM SVC / Storwize V3700 / V7000 devices * 1918 ps: new option for checking the age of a process (on Linux)... * 1920 df: Linux filesystem check now supports displaying data reserved for root... + * 1675 esx_vsphere_hostsystem.cpu_util_cluster: Averaged CPU utilization of all cluster nodes... * 1457 FIX: logins: new check renamed from "users" check... NOTE: Please refer to the migration notes! * 1762 FIX: lnx_thermal: Now ignoring trip points with level 0... diff --git a/checkman/esx_vsphere_hostsystem.cpu_util_cluster b/checkman/esx_vsphere_hostsystem.cpu_util_cluster new file mode 100644 index 0000000..2d34c32 --- /dev/null +++ b/checkman/esx_vsphere_hostsystem.cpu_util_cluster @@ -0,0 +1,32 @@ +title: VMWare ESX host systems: CPU utilization of clusters +agents: vsphere +catalog: os/kernel +license: GPL +distribution: check_mk +description: + This check measures the averaged CPU utilization for a cluster of ESX Host systems. + You can configure multiple WARN/CRIT levels depending on the number of nodes in this cluster. + This check is only applicable for clusters and cannot be inventorized. + You need to create a manual check for it in WATO and assign it to a cluster host. + +examples: + static_checks['cpu_utilization_cluster'] = [ + ( ('esx_vsphere_hostsystem.cpu_util_cluster', None, + [(2, (60.0, 70.0)), + (5, (70.0, 80.0)), + (7, (85.0, 90.0)) + ]), [], ['esx_cluster'] ), + ] + static_checks['cpu_utilization_cluster'] + +perfdata: + One value: The current usage in percent - ranging from 0 to 100. + The "maximum" value is not 100, but the number of CPU threads. This + case be used for scaling the graph in terms of the number of used CPU threads. + +[parameters] +parameters (list): A list of tuples where you can define the number of + minimum required nodes and the corresponding levels. + + One Element: ( number_of_nodes(int), ( WARN(float), CRIT(float) ) ) + + Please refer example. diff --git a/checkman/esx_vsphere_hostsystem.mem_usage_cluster b/checkman/esx_vsphere_hostsystem.mem_usage_cluster index 1d019b3..4ab8bd0 100644 --- a/checkman/esx_vsphere_hostsystem.mem_usage_cluster +++ b/checkman/esx_vsphere_hostsystem.mem_usage_cluster @@ -17,11 +17,6 @@ examples: ]), [], ['esx_cluster'] ), ] + static_checks['mem_cluster'] - - -inventory: - Not applicable - perfdata: One value: the current total usage in bytes. diff --git a/checks/esx_vsphere_hostsystem b/checks/esx_vsphere_hostsystem index 7b9a57d..be09a49 100644 --- a/checks/esx_vsphere_hostsystem +++ b/checks/esx_vsphere_hostsystem @@ -148,7 +148,83 @@ check_info['esx_vsphere_hostsystem.mem_usage_cluster'] = { } +#. +# .--CPU-Cluster---------------------------------------------------------. +# | ____ ____ _ _ ____ _ _ | +# | / ___| _ \| | | | / ___| |_ _ ___| |_ ___ _ __ | +# | | | | |_) | | | |_____| | | | | | / __| __/ _ \ '__| | +# | | |___| __/| |_| |_____| |___| | |_| \__ \ || __/ | | +# | \____|_| \___/ \____|_|\__,_|___/\__\___|_| | +# | | +# +----------------------------------------------------------------------+ + + +def check_esx_vsphere_hostsystem_cpu_util_cluster(item, params, info): + current_node = {} + def get_node_usage(node): + num_sockets = int(node['hardware.cpuInfo.numCpuPackages']) + num_cores = int(node['hardware.cpuInfo.numCpuCores']) + num_threads = int(node['hardware.cpuInfo.numCpuThreads']) + used_mhz = float(node['summary.quickStats.overallCpuUsage']) + mhz_per_core = float(node['hardware.cpuInfo.hz']) / 1024.0 / 1024.0 + total_mhz = mhz_per_core * num_cores + return used_mhz, total_mhz, num_threads + + overall_used = [] + overall_total = [] + overall_threads = [] + for line in info: + if line[0] in [ "hardware.cpuInfo.numCpuPackages", + "hardware.cpuInfo.numCpuCores", + "hardware.cpuInfo.numCpuThreads", + "summary.quickStats.overallCpuUsage", + "hardware.cpuInfo.hz" ]: + current_node[line[0]] = line[1] + if len(current_node) == 5: # 5 keys -> node complete + node_used, node_total, node_threads = get_node_usage(current_node) + overall_used.append(node_used) + overall_total.append(node_total) + overall_threads.append(node_threads) + current_node = {} + + + sum_used = sum(overall_used) + sum_total = sum(overall_total) + sum_threads = sum(overall_threads) + usage = sum_used / sum_total * 100 + node_count = len(overall_used) + + # Convert legacy parameters + this_time = time.time() + + + sorted_params = sorted(params, reverse = True) + for count, levels in sorted_params: + if node_count >= count: + state, infotext, perfdata = check_cpu_util(usage, levels) + break + else: + state, infotext, perfdata = check_cpu_util(usage, None) + yield 0, "%d Nodes" % node_count + yield 0, "%.2fGHz/%.2fGHz" % (sum_used / 1024.0, sum_total / 1024.0) + yield 0, "%d threads" % sum_threads + + # put number of threads as MAX value for first perf-data. This + # is needed by the PNP template. + perfdata_cpu = list(perfdata[0]) + perfdata_cpu[-1] = sum_threads + perfdata = [ tuple(perfdata_cpu) ] + perfdata[1:] + yield state, infotext, perfdata + + +check_info['esx_vsphere_hostsystem.cpu_util_cluster'] = { + "check_function" : check_esx_vsphere_hostsystem_cpu_util_cluster, + "service_description" : "CPU utilization", + "group" : "cpu_utilization_cluster", + "has_perfdata" : True, + "includes" : [ "cpu_util.include" ], +} # .--Memory--------------------------------------------------------------. # | __ __ | # | | \/ | ___ _ __ ___ ___ _ __ _ _ | diff --git a/web/plugins/wato/check_parameters.py b/web/plugins/wato/check_parameters.py index 7455602..d3e87f0 100644 --- a/web/plugins/wato/check_parameters.py +++ b/web/plugins/wato/check_parameters.py @@ -2998,6 +2998,35 @@ register_check_parameters( ) register_check_parameters( + subgroup_networking, + "cpu_utilization_cluster", + _("CPU Utilization of Clusters"), + ListOf( + Tuple( + elements = [ + Integer(title = _("Equal or more than"), unit = _("nodes")), + Tuple( + elements = [ + Percentage(title = _("Warning at a utilization of"), default_value = 90.0), + Percentage(title = _("Critical at a utilization of"), default_value = 95.0) + ], + title = _("Alert on too high CPU utilization"), + ) + ] + ), + help = _("Configure levels for averaged CPU utilization depending on number of cluster nodes. " + "The CPU utilization sums up the percentages of CPU time that is used " + "for user processes and kernel routines over all available cores within " + "the last check interval. The possible range is from 0% to 100%"), + title = _("Memory Usage"), + add_label = _("Add limits") + ), + None, + "first", + False +) + +register_check_parameters( subgroup_os, "esx_host_memory", _("Main memory usage of ESX host system"),