def test_classify_process(self):
mc = MetricClassifier(goodmetricdefs)
self.assertEqual(mc.classify_metrics(samplemetrics), samplemetricresults)
self.assertEqual(mc.worst_metric(samplemetrics), ('d', 3))
self.assertEqual(mc.return_code(samplemetrics), 3)
self.assertEqual(mc.return_code(samplemetrics, unknown_as_critical=True), 2)
class NTPAlerter(object):
def __init__(self, checks, objs):
self.checks = checks
self.objs = objs
self.mc = MetricClassifier(_metricdefs)
def collectmetrics(self, debug):
"""
Get metrics from each registered metric source and add all relevant aliases.
"""
self.metrics = {}
for o in self.objs:
self.metrics.update(self.objs[o].getmetrics())
if debug:
pprint.pprint(self.metrics)
metrics.addaliases(self.metrics, _aliases)
if 'proc' in self.checks:
self.checks.append('runtime')
if 'trace' in self.checks:
self.checks.append('tracehosts')
self.checks.append('traceloops')
if 'vars' in self.checks and 'offset' not in self.checks:
self.checks.append('sysoffset')
def custom_message(self, metric, result):
"""
Special cases for message formats
"""
if metric == 'runtime':
return self.custom_message_runtime(result)
elif metric == 'sync':
return self.custom_message_sync(result)
elif metric == 'tracehosts':
return self.custom_message_tracehosts(result)
elif metric == 'traceloops':
return self.custom_message_traceloops(result)
return None
def custom_message_runtime(self, result):
proc = self.objs['proc']
if result == 'CRITICAL':
return '%s: No NTP process could be found. Please check that an NTP server is installed and running.' % (
result, )
elif result == 'WARNING':
return 'OK: %s has only been running %d seconds' % (
proc.name, proc.getruntime())
elif result == 'OK':
return '%s: %s has been running %d seconds' % (result, proc.name,
proc.getruntime())
return None
def custom_message_sync(self, result):
if result == 'CRITICAL':
return '%s: No sync peer selected' % (result, )
elif result == 'OK':
return '%s: Time is in sync with %s' % (
result, self.objs['peers'].syncpeer())
return None
def custom_message_traceloops(self, result):
if result == 'CRITICAL':
return '%s: Trace loop detected at host %s' % (
result, self.objs['trace'].loophost)
elif result == 'OK':
return '%s: Trace detected no loops' % (result, )
return None
def custom_message_tracehosts(self, result):
trace = self.objs['trace']
return '%s: %d hosts detected in trace: %s' % (
result, trace.results['tracehosts'], ", ".join(trace.hostlist))
def alert_collectd(self, hostname, interval):
"""
Produce collectd output for the metrics
"""
self.collectmetrics(False)
self.mc.classify_metrics(self.metrics)
(m, rc) = self.mc.worst_metric(self.checks)
self.metrics['result'] = self.return_code()
for metric in sorted(_collectdtypes.keys()):
if metric in _collectdtypes and metric in self.metrics:
print('PUTVAL "%s/ntpmon-%s" interval=%d N:%.9f' % (
hostname,
_collectdtypes[metric],
interval,
self.metrics[metric],
))
if sys.stdout.isatty():
# we're outputting to a terminal; must be test mode
print('')
else:
# flush standard output to ensure metrics are sent to collectd immediately
sys.stdout.flush()
def alert_nagios(self, debug):
"""
Produce nagios output for the metrics
"""
self.collectmetrics(debug)
results = self.mc.classify_metrics(self.metrics)
msgs = {}
for metric in self.checks:
if metric in results:
msgs[metric] = self.custom_message(metric, results[metric])
if msgs[metric] is None:
msgs[metric] = self.mc.message(metric, _formats[metric][0],
_formats[metric][1])
if debug:
for m in msgs:
print(msgs[m])
else:
(m, rc) = self.mc.worst_metric(self.checks)
self.metrics['result'] = self.return_code()
print("%s | %s" % (msgs[m], self.report()))
def report(self):
"""
Report metric values.
"""
items = []
for m in sorted(_aliases.keys()):
if m in self.metrics:
if _formats[m] is None:
fmt = 'f'
else:
fmt = _formats[m][1] if _formats[m][1] != '%' else 'f'
val = self.mc.fmtstr(fmt) % self.metrics[m]
items.append("%s=%s" % (m, val))
else:
items.append("%s=" % (m, ))
return " ".join(items)
def return_code(self):
"""
Don't return anything other than OK until ntpd has been running for
at least enough time for 8 polling intervals of 64 seconds each. This
prevents false positives due to ntpd restarts or short-lived VMs.
"""
if 'runtime' in self.mc.results and self.mc.results[
'runtime'] == 'WARNING':
return 0
else:
return self.mc.return_code(self.checks)
def test_classify_multiple_metrics(self):
mc = MetricClassifier(goodmetricdefs)
self.assertEqual(mc.classify_metrics(samplemetrics), samplemetricresults)
class NTPAlerter(object):
def __init__(self, checks, objs):
self.checks = checks
self.objs = objs
self.mc = MetricClassifier(_metricdefs)
def collectmetrics(self, debug):
"""
Get metrics from each registered metric source and add all relevant aliases.
"""
self.metrics = {}
for o in self.objs:
self.metrics.update(self.objs[o].getmetrics())
if debug:
pprint.pprint(self.metrics)
metrics.addaliases(self.metrics, _aliases)
if 'proc' in self.checks:
self.checks.append('runtime')
if 'trace' in self.checks:
self.checks.append('tracehosts')
self.checks.append('traceloops')
if 'vars' in self.checks and 'offset' not in self.checks:
self.checks.append('sysoffset')
def custom_message(self, metric, result):
"""
Special cases for message formats
"""
if metric == 'runtime':
return self.custom_message_runtime(result)
elif metric == 'sync':
return self.custom_message_sync(result)
elif metric == 'tracehosts':
return self.custom_message_tracehosts(result)
elif metric == 'traceloops':
return self.custom_message_traceloops(result)
return None
def custom_message_runtime(self, result):
proc = self.objs['proc']
if result == 'CRITICAL':
return '%s: No NTP process could be found. Please check that an NTP server is installed and running.' % (result,)
elif result == 'WARNING':
return 'OK: %s has only been running %d seconds' % (proc.name, proc.getruntime())
elif result == 'OK':
return '%s: %s has been running %d seconds' % (result, proc.name, proc.getruntime())
return None
def custom_message_sync(self, result):
if result == 'CRITICAL':
return '%s: No sync peer selected' % (result,)
elif result == 'OK':
return '%s: Time is in sync with %s' % (result, self.objs['peers'].syncpeer())
return None
def custom_message_traceloops(self, result):
if result == 'CRITICAL':
return '%s: Trace loop detected at host %s' % (result, self.objs['trace'].loophost)
elif result == 'OK':
return '%s: Trace detected no loops' % (result,)
return None
def custom_message_tracehosts(self, result):
trace = self.objs['trace']
return '%s: %d hosts detected in trace: %s' % (
result,
trace.results['tracehosts'],
", ".join(trace.hostlist)
)
def alert_collectd(self, hostname, interval):
"""
Produce collectd output for the metrics
"""
self.collectmetrics(False)
self.mc.classify_metrics(self.metrics)
(m, rc) = self.mc.worst_metric(self.checks)
self.metrics['result'] = self.return_code()
for metric in sorted(_collectdtypes.keys()):
if metric in _collectdtypes and metric in self.metrics:
print('PUTVAL "%s/ntpmon-%s" interval=%d N:%.9f' % (
hostname,
_collectdtypes[metric],
interval,
self.metrics[metric],
))
if sys.stdout.isatty():
# we're outputting to a terminal; must be test mode
print('')
else:
# flush standard output to ensure metrics are sent to collectd immediately
sys.stdout.flush()
def alert_nagios(self, debug):
"""
Produce nagios output for the metrics
"""
self.collectmetrics(debug)
results = self.mc.classify_metrics(self.metrics)
msgs = {}
for metric in self.checks:
if metric in results:
msgs[metric] = self.custom_message(metric, results[metric])
if msgs[metric] is None:
msgs[metric] = self.mc.message(metric, _formats[metric][0], _formats[metric][1])
if debug:
for m in msgs:
print(msgs[m])
else:
(m, rc) = self.mc.worst_metric(self.checks)
self.metrics['result'] = self.return_code()
print("%s | %s" % (msgs[m], self.report()))
def report(self):
"""
Report metric values.
"""
items = []
for m in sorted(_aliases.keys()):
if m in self.metrics:
if _formats[m] is None:
fmt = 'f'
else:
fmt = _formats[m][1] if _formats[m][1] != '%' else 'f'
val = self.mc.fmtstr(fmt) % self.metrics[m]
items.append("%s=%s" % (m, val))
else:
items.append("%s=" % (m,))
return " ".join(items)
def return_code(self):
"""
Don't return anything other than OK until ntpd has been running for
at least enough time for 8 polling intervals of 64 seconds each. This
prevents false positives due to ntpd restarts or short-lived VMs.
"""
if 'runtime' in self.mc.results and self.mc.results['runtime'] == 'WARNING':
return 0
else:
return self.mc.return_code(self.checks)