Module: check_mk Branch: master Commit: eaa67d79f3c0f46313e1300807aed7b9ea9eff64 URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=eaa67d79f3c0f4… Author: Mathias Kettner <mk(a)mathias-kettner.de> Date: Sun Nov 14 15:56:36 2010 +0100 kernel: convert perfdata to rates, perfometer --- ChangeLog | 5 +++++ checkman/kernel | 11 ++++++----- checks/kernel | 20 ++++++++++++++------ pnp-templates/check_mk-kernel.php | 8 ++++---- web/plugins/perfometer/check_mk.py | 6 ++++++ 5 files changed, 35 insertions(+), 15 deletions(-) diff --git a/ChangeLog b/ChangeLog index 197dacc..1395a16 100644 --- a/ChangeLog +++ b/ChangeLog @@ -17,6 +17,11 @@ * Renamed check functions of imm_health check from test_imm to imm_health to have valid function and check names. Please remove remove from inventory and re-inventory those checks. + * Check kernel: converted performance data from counters to rates. This + fixes RRD problems (spikes) on reboots and also allows better access + to the peformance data for the Perf-O-Meters. Also changed service + descriptions. You need to reinventurize the kernel checks. You old + RRDs will not be deleted, new ones will be created. Core, Setup, etc.: * Improve error handling: if hosts are monitored with SNMP *and* TCP, diff --git a/checkman/kernel b/checkman/kernel index 6270609..e1308d4 100644 --- a/checkman/kernel +++ b/checkman/kernel @@ -16,21 +16,23 @@ description: be checked. item: - The name of the kernel counter as found in {/proc/stat} or + {"Context Switches"}, {"Process Creations"}, {"Major Page Faults"} or the name + of the kernel counter in question as found in {/proc/stat} or {/proc/vmstat}. examples: # Monitor major page faults, set levels to 200 and 400 per second # Do this on all hosts with the tag "lnx" checks += [ - ( ["lnx"], ALL_HOSTS, "kernel", "pgmajfault", (200, 400) ) + ( ["lnx"], ALL_HOSTS, "kernel", "Major Page Faults", (200, 400) ) ] perfdata: One variable: the current value of the counter. inventory: - Per default one check for {pgmajfault}, {ctxt} and {processes} + Per default one check for {"Major Page Faults"}, + {"Context Switches"} and {"Process Creations"} will be created on each host providing that data. You can override this list globally by setting {inventory_kernel_counters} to a list of strings with those counters that should be inventorized. @@ -39,7 +41,7 @@ inventory: examples: # Make inventory create only checks for pgmajfault - inventory_kernel_counters = [ "pgmajfault" ] + inventory_kernel_counters = [ "pgmajfault", "pgpgin", "pgpgout" ] [parameters] warning (int): The rate per second which triggers a warning. A value of {None} disables the level. @@ -51,4 +53,3 @@ inventory_kernel_counters (list of strings): List of performance counters the in kernel_default_levels (int, int): Default levels for newly inventorized checks. Default is {(None, None)}, which disables the levels and makes the check always OK. - diff --git a/checks/kernel b/checks/kernel index fb6aa9b..85acd25 100644 --- a/checks/kernel +++ b/checks/kernel @@ -28,12 +28,19 @@ inventory_kernel_counters = [ "pgmajfault", "ctxt", "processes" ] kernel_default_levels = (None, None) +kernel_counter_names = { + "ctxt" : "Context Switches", + "processes" : "Process Creations", + "pgmajfault" : "Major Page Faults", +} + def inventory_kernel(checktype, info): inventory = [] for counter in inventory_kernel_counters: hits = [ line[0] for line in info[1:] if line[0] == counter ] if len(hits) == 1: - inventory.append( (counter, "kernel_default_levels") ) + countername = kernel_counter_names.get(counter, counter) + inventory.append( (countername, "kernel_default_levels") ) return inventory @@ -41,21 +48,22 @@ def inventory_kernel(checktype, info): def check_kernel(item, params, info): this_time = int(info[0][0]) - hits = [ line[1] for line in info[1:] if line[0] == item ] + hits = [ (line[0], line[1]) for line in info[1:] if line[0] == item or kernel_counter_names.get(line[0], line[0]) == item ] if len(hits) == 0: return (3, "UNKNOWN - item '%s' not found in agent output" % item) elif len(hits) > 1: return (3, "UNKNOWN - item '%s' not unique (found %d times)" % (item, len(hits))) - this_val = int(hits[0]) + counter = hits[0][0] + this_val = int(hits[0][1]) timedif, per_sec = get_counter("kernel." + item, this_time, this_val) infotext = " - %.0f/s in last %d secs" % (per_sec, timedif) + if params == None: - perfdata = [ (item, "%dc" % this_val, "", "", 0 ) ] - return (0, "OK" + infotext, perfdata) + return (0, "OK" + infotext, [ (counter, per_sec) ]) warn, crit = params - perfdata = [ (item, "%dc" % this_val, warn, crit, 2) ] + perfdata = [ (counter, per_sec, warn, crit) ] if warn == None and crit != None: infotext += " (critical at %.0f/s)" % crit elif warn != None and crit == None: diff --git a/pnp-templates/check_mk-kernel.php b/pnp-templates/check_mk-kernel.php index c6e9052..d2bf69d 100644 --- a/pnp-templates/check_mk-kernel.php +++ b/pnp-templates/check_mk-kernel.php @@ -24,7 +24,7 @@ # Boston, MA 02110-1301 USA. $subtype = substr($servicedesc, 7); -if ($subtype == "pgmajfault") { +if ($subtype == "pgmajfault" || $subtype == "Major_Page_Faults") { $title = "Major Page Faults"; $vertical = "faults / sec"; $format = "%5.1lf/s"; @@ -32,7 +32,7 @@ if ($subtype == "pgmajfault") { $color = "20ff80"; $line = "10a040"; } -else if ($subtype == "ctxt") { +else if ($subtype == "ctxt" || $subtype == "Context_Switches") { $title = "Context Switches"; $vertical = "switches / sec"; $format = "%5.1lf/s"; @@ -40,7 +40,7 @@ else if ($subtype == "ctxt") { $color = "80ff20"; $line = "40a010"; } -else if ($subtype == "processes") { +else if ($subtype == "processes" || $subtype == "Process_Creations") { $title = "Process creation"; $vertical = "new processes / sec"; $format = "%5.1lf/s"; @@ -57,7 +57,7 @@ else { $line = "90a010"; } -$opt[1] = " --vertical-label \"$vertical\" -X0 -l 0 -u $upto --title \"$title\" "; +$opt[1] = " --vertical-label \"$vertical\" -X0 -l 0 -u $upto --title \"$hostname: $title\" "; $def[1] = "DEF:var1=$RRDFILE[1]:$DS[1]:MAX "; $def[1] .= "AREA:var1#$color:\"$title\:\" "; diff --git a/web/plugins/perfometer/check_mk.py b/web/plugins/perfometer/check_mk.py index ce5cd4e..28cdc71 100644 --- a/web/plugins/perfometer/check_mk.py +++ b/web/plugins/perfometer/check_mk.py @@ -120,6 +120,12 @@ def perfometer_check_mk_cpu_threads(row, check_command, perf_data): perfometers["check_mk-cpu.threads"] = perfometer_check_mk_cpu_threads +def perfometer_check_mk_kernel(row, check_command, perf_data): + rate = float(perf_data[0][1]) + return "%.1f/s" % rate, perfometer_logarithmic(rate, 1000, 2, "#da6") + +perfometers["check_mk-kernel"] = perfometer_check_mk_kernel + def perfometer_check_mk_cpu_loads(row, check_command, perf_data): color = { 0: "#68f", 1: "#ff2", 2: "#f22", 3: "#fa2" }[row["service_state"]]