def run(self):
data = {}
for profile in self.config:
for region in self.config[profile]:
for db in self.config[profile][region]:
if (profile == 'default'):
profile = None
rds = BotoRDS(region=region,
profile=profile,
identifier=db)
for metric in self.metrics.keys():
if metric == 'FreeableMemory':
info = rds.get_info()
try:
memory = self.db_classes[
info.instance_class] * 1024**3
except IndexError:
print 'Unknown DB instance class "%s"' % info.instance_class
sys.exit(1)
data['{0}_{1}_{2}'.format(
region, db, 'UsedMemory')] = memory - stats
data['{0}_{1}_{2}'.format(region, db,
'TotalMemory')] = memory
elif metric == 'FreeStorageSpace':
info = rds.get_info()
storage = float(info.allocated_storage) * 1024**3
data['{0}_{1}_{2}'.format(
region, db, 'UsedDiskUsage')] = storage - stats
data['{0}_{1}_{2}'.format(
region, db, 'TotalDiskUsage')] = storage
else:
stats = rds.get_metric(metric)
data['{0}_{1}_{2}'.format(region, db,
metric)] = stats
return data
def fetch_stats(self, data, rds):
for metric, values in self.metrics.items():
try:
stats = rds.get_metric(metric)
inst = rds.identifier
if metric in self.byte_related:
# formatting to megabytes
stats = stats / 10**6
if metric == 'FreeableMemory':
info = rds.get_info()
try:
memory = self.db_classes[info.instance_class] * 1000
used_mem = memory - stats
mem_name = self.metrics[metric]
data['{0}_{1}'.format(inst, 'used_memory')] = used_mem
data['{0}_{1}'.format(inst, 'total_memory')] = memory
data['{0}_{1}'.format(inst, mem_name)] = stats
except IndexError as e:
msg = 'RDS: Unknown DB instance class "{}"'
self.checks_logger.error(
msg.format(info.instance_class)
)
elif metric == 'FreeStorageSpace':
info = rds.get_info()
storage = float(info.allocated_storage) * 1000
used = storage - stats
data['{0}_{1}'.format(inst, self.metrics[metric])] = stats
data['{0}_{1}'.format(inst, 'used_diskusage')] = used
data['{0}_{1}'.format(inst, 'total_diskusage')] = storage
elif metric in self.byte_related:
data['{0}_{1}'.format(inst, self.metrics[metric])] = stats
else:
data['{0}_{1}'.format(inst, self.metrics[metric])] = stats
except NoMetricError as e:
msg = 'RDS: {} was not available for {}'
self.checks_logger.info(msg.format(metric, rds.identifier))
def main():
"""Main function"""
global options
short_status = {
OK: 'OK',
WARNING: 'WARN',
CRITICAL: 'CRIT',
UNKNOWN: 'UNK'
}
# DB instance classes as listed on
# http://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/Concepts.DBInstanceClass.html
db_classes = {
'db.t1.micro': 0.615,
'db.m1.small': 1.7,
'db.m1.medium': 3.75,
'db.m1.large': 7.5,
'db.m1.xlarge': 15,
'db.m4.large': 8,
'db.m4.xlarge': 16,
'db.m4.2xlarge': 32,
'db.m4.4xlarge': 64,
'db.m4.10xlarge': 160,
'db.r3.large': 15,
'db.r3.xlarge': 30.5,
'db.r3.2xlarge': 61,
'db.r3.4xlarge': 122,
'db.r3.8xlarge': 244,
'db.t2.micro': 1,
'db.t2.small': 2,
'db.t2.medium': 4,
'db.t2.large': 8,
'db.m3.medium': 3.75,
'db.m3.large': 7.5,
'db.m3.xlarge': 15,
'db.m3.2xlarge': 30,
'db.m2.xlarge': 17.1,
'db.m2.2xlarge': 34.2,
'db.m2.4xlarge': 68.4,
'db.cr1.8xlarge': 244,
}
# RDS metrics http://docs.aws.amazon.com/AmazonCloudWatch/latest/DeveloperGuide/rds-metricscollected.html
metrics = {
'status': 'RDS availability',
'load': 'CPUUtilization',
'memory': 'FreeableMemory',
'storage': 'FreeStorageSpace'
}
units = ('percent', 'GB')
# Parse options
parser = optparse.OptionParser()
parser.add_option('-l', '--list', help='list of all DB instances',
action='store_true', default=False, dest='db_list')
parser.add_option('-n', '--profile', default=None,
help='AWS profile from ~/.boto or /etc/boto.cfg. Default: None, fallbacks to "[Credentials]".')
parser.add_option('-r', '--region', default='us-east-1',
help='AWS region. Default: us-east-1. If set to "all", we try to detect the instance region '
'across all of them, note this will be slower than if you specify the region explicitly.')
parser.add_option('-i', '--ident', help='DB instance identifier')
parser.add_option('-p', '--print', help='print status and other details for a given DB instance',
action='store_true', default=False, dest='printinfo')
parser.add_option('-m', '--metric', help='metric to check: [%s]' % ', '.join(metrics.keys()))
parser.add_option('-w', '--warn', help='warning threshold')
parser.add_option('-c', '--crit', help='critical threshold')
parser.add_option('-u', '--unit', help='unit of thresholds for "storage" and "memory" metrics: [%s]. '
'Default: percent' % ', '.join(units), default='percent')
parser.add_option('-t', '--time', help='time period in minutes to query. Default: 5',
type='int', default=5)
parser.add_option('-a', '--avg', help='time average in minutes to request. Default: 1',
type='int', default=1)
parser.add_option('-d', '--debug', help='enable debug output',
action='store_true', default=False)
options, _ = parser.parse_args()
if options.debug:
boto.set_stream_logger('boto')
rds = RDS(region=options.region, profile=options.profile, identifier=options.ident)
# Check args
if len(sys.argv) == 1:
parser.print_help()
sys.exit()
elif options.db_list:
info = rds.get_list()
print 'List of all DB instances in %s region(s):' % (options.region,)
pprint.pprint(info)
sys.exit()
elif not options.ident:
parser.print_help()
parser.error('DB identifier is not set.')
elif options.printinfo:
info = rds.get_info()
if info:
pprint.pprint(vars(info))
else:
print 'No DB instance "%s" found on your AWS account and %s region(s).' % (options.ident, options.region)
sys.exit()
elif not options.metric or options.metric not in metrics.keys():
parser.print_help()
parser.error('Metric is not set or not valid.')
elif not options.warn and options.metric != 'status':
parser.print_help()
parser.error('Warning threshold is not set.')
elif not options.crit and options.metric != 'status':
parser.print_help()
parser.error('Critical threshold is not set.')
elif options.avg <= 0 and options.metric != 'status':
parser.print_help()
parser.error('Average must be greater than zero.')
elif options.time <= 0 and options.metric != 'status':
parser.print_help()
parser.error('Time must be greater than zero.')
now = datetime.datetime.utcnow()
status = None
note = ''
perf_data = None
# RDS Status
if options.metric == 'status':
info = rds.get_info()
if not info:
status = UNKNOWN
note = 'Unable to get RDS instance'
else:
status = OK
try:
version = info.EngineVersion
except:
version = info.engine_version
note = '%s %s. Status: %s' % (info.engine, version, info.status)
# RDS Load Average
elif options.metric == 'load':
# Check thresholds
try:
warns = [float(x) for x in options.warn.split(',')]
crits = [float(x) for x in options.crit.split(',')]
fail = len(warns) + len(crits)
except:
fail = 0
if fail != 6:
parser.error('Warning and critical thresholds should be 3 comma separated numbers, e.g. 20,15,10')
loads = []
fail = False
j = 0
perf_data = []
for i in [1, 5, 15]:
if i == 1:
# Some stats are delaying to update on CloudWatch.
# Let's pick a few points for 1-min load avg and get the last point.
points = 5
else:
points = i
load = rds.get_metric(metrics[options.metric], now - datetime.timedelta(seconds=points * 60), now, i * 60)
if not load:
status = UNKNOWN
note = 'Unable to get RDS statistics'
perf_data = None
break
loads.append(str(load))
perf_data.append('load%s=%s;%s;%s;0;100' % (i, load, warns[j], crits[j]))
# Compare thresholds
if not fail:
if warns[j] > crits[j]:
parser.error('Parameter inconsistency: warning threshold is greater than critical.')
elif load >= crits[j]:
status = CRITICAL
fail = True
elif load >= warns[j]:
status = WARNING
j = j + 1
if status != UNKNOWN:
if status is None:
status = OK
note = 'Load average: %s%%' % '%, '.join(loads)
perf_data = ' '.join(perf_data)
# RDS Free Storage
# RDS Free Memory
elif options.metric in ['storage', 'memory']:
# Check thresholds
try:
warn = float(options.warn)
crit = float(options.crit)
except:
parser.error('Warning and critical thresholds should be integers.')
if crit > warn:
parser.error('Parameter inconsistency: critical threshold is greater than warning.')
if options.unit not in units:
parser.print_help()
parser.error('Unit is not valid.')
info = rds.get_info()
free = rds.get_metric(metrics[options.metric], now - datetime.timedelta(seconds=options.time * 60),
now, options.avg * 60)
if not info or not free:
status = UNKNOWN
note = 'Unable to get RDS details and statistics'
else:
if options.metric == 'storage':
storage = float(info.allocated_storage)
elif options.metric == 'memory':
try:
storage = db_classes[info.instance_class]
except:
print 'Unknown DB instance class "%s"' % info.instance_class
sys.exit(CRITICAL)
free = '%.2f' % (free / 1024 ** 3)
free_pct = '%.2f' % (float(free) / storage * 100)
if options.unit == 'percent':
val = float(free_pct)
val_max = 100
elif options.unit == 'GB':
val = float(free)
val_max = storage
# Compare thresholds
if val <= crit:
status = CRITICAL
elif val <= warn:
status = WARNING
if status is None:
status = OK
note = 'Free %s: %s GB (%.0f%%) of %s GB' % (options.metric, free, float(free_pct), storage)
perf_data = 'free_%s=%s;%s;%s;0;%s' % (options.metric, val, warn, crit, val_max)
# Final output
if status != UNKNOWN and perf_data:
print '%s %s | %s' % (short_status[status], note, perf_data)
else:
print '%s %s' % (short_status[status], note)
sys.exit(status)
def main():
"""Main function"""
global options
# DB instance classes as listed on
# http://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/Concepts.DBInstanceClass.html
db_classes = {
'db.t1.micro': 0.615,
'db.m1.small': 1.7,
'db.m1.medium': 3.75,
'db.m1.large': 7.5,
'db.m1.xlarge': 15,
'db.m4.large': 8,
'db.m4.xlarge': 16,
'db.m4.2xlarge': 32,
'db.m4.4xlarge': 64,
'db.m4.10xlarge': 160,
'db.r3.large': 15,
'db.r3.xlarge': 30.5,
'db.r3.2xlarge': 61,
'db.r3.4xlarge': 122,
'db.r3.8xlarge': 244,
'db.t2.micro': 1,
'db.t2.small': 2,
'db.t2.medium': 4,
'db.t2.large': 8,
'db.m3.medium': 3.75,
'db.m3.large': 7.5,
'db.m3.xlarge': 15,
'db.m3.2xlarge': 30,
'db.m2.xlarge': 17.1,
'db.m2.2xlarge': 34.2,
'db.m2.4xlarge': 68.4,
'db.cr1.8xlarge': 244,
}
# RDS metrics http://docs.aws.amazon.com/AmazonCloudWatch/latest/DeveloperGuide/rds-metricscollected.html
metrics = {
'BinLogDiskUsage': 'binlog_disk_usage', # The amount of disk space occupied by binary logs on the master. Units: Bytes
'CPUUtilization': 'utilization', # The percentage of CPU utilization. Units: Percent
'DatabaseConnections': 'connections', # The number of database connections in use. Units: Count
'DiskQueueDepth': 'disk_queue_depth', # The number of outstanding IOs (read/write requests) waiting to access the disk. Units: Count
'ReplicaLag': 'replica_lag', # The amount of time a Read Replica DB Instance lags behind the source DB Instance. Units: Seconds
'SwapUsage': 'swap_usage', # The amount of swap space used on the DB Instance. Units: Bytes
'FreeableMemory': 'used_memory', # The amount of available random access memory. Units: Bytes
'FreeStorageSpace': 'used_space', # The amount of available storage space. Units: Bytes
'ReadIOPS': 'read_iops', # The average number of disk I/O operations per second. Units: Count/Second
'WriteIOPS': 'write_iops', # The average number of disk I/O operations per second. Units: Count/Second
'ReadLatency': 'read_latency', # The average amount of time taken per disk I/O operation. Units: Seconds
'WriteLatency': 'write_latency', # The average amount of time taken per disk I/O operation. Units: Seconds
'ReadThroughput': 'read_throughput', # The average number of bytes read from disk per second. Units: Bytes/Second
'WriteThroughput': 'write_throughput', # The average number of bytes written to disk per second. Units: Bytes/Second
}
# Parse options
parser = optparse.OptionParser()
parser.add_option('-l', '--list', help='list DB instances',
action='store_true', default=False, dest='db_list')
parser.add_option('-n', '--profile', default=None,
help='AWS profile from ~/.boto or /etc/boto.cfg. Default: None, fallbacks to "[Credentials]".')
parser.add_option('-r', '--region', default='us-east-1',
help='AWS region. Default: us-east-1. If set to "all", we try to detect the instance region '
'across all of them, note this will be slower than if you specify the region explicitly.')
parser.add_option('-i', '--ident', help='DB instance identifier')
parser.add_option('-p', '--print', help='print status and other details for a given DB instance',
action='store_true', default=False, dest='printinfo')
parser.add_option('-m', '--metric', help='metrics to retrive separated by comma: [%s]' % ', '.join(metrics.keys()))
parser.add_option('-d', '--debug', help='enable debugging',
action='store_true', default=False)
options, _ = parser.parse_args()
# Strip a prefix _ which is sent by Cacti, so an empty argument is interpreted correctly.
# Than set defaults if argument is supposed to be empty.
options.region = options.region.lstrip('_')
options.profile = options.profile.lstrip('_')
if not options.region:
options.region = 'us-east-1'
if not options.profile:
options.profile = None
if options.debug:
boto.set_stream_logger('boto')
rds = RDS(region=options.region, profile=options.profile, identifier=options.ident)
# Check args
if len(sys.argv) == 1:
parser.print_help()
sys.exit()
elif options.db_list:
info = rds.get_list()
print 'List of all DB instances in %s region(s):' % (options.region,)
pprint.pprint(info)
sys.exit()
elif not options.ident:
parser.print_help()
parser.error('DB identifier is not set.')
elif options.printinfo:
info = rds.get_info()
pprint.pprint(vars(info))
sys.exit()
elif not options.metric:
parser.print_help()
parser.error('Metric is not set.')
selected_metrics = options.metric.split(',')
for metric in selected_metrics:
if metric not in metrics.keys():
parser.print_help()
parser.error('Invalid metric.')
# Do not remove the empty lines in the start and end of this docstring
perl_magic_vars = """
# Define the variables to output. I use shortened variable names so maybe
# it'll all fit in 1024 bytes for Cactid and Spine's benefit. Strings must
# have some non-hex characters (non a-f0-9) to avoid a Cacti bug. This list
# must come right after the word MAGIC_VARS_DEFINITIONS. The Perl script
# parses it and uses it as a Perl variable.
$keys = array(
'binlog_disk_usage' => 'gg',
'utilization' => 'gh',
'connections' => 'gi',
'disk_queue_depth' => 'gj',
'replica_lag' => 'gk',
'swap_usage' => 'gl',
'used_memory' => 'gm',
'total_memory' => 'gn',
'used_space' => 'go',
'total_space' => 'gp',
'read_iops' => 'gq',
'write_iops' => 'gr',
'read_latency' => 'gs',
'write_latency' => 'gt',
'read_throughput' => 'gu',
'write_throughput' => 'gv',
);
"""
output = dict()
for row in perl_magic_vars.split('\n'):
if row.find('=>') >= 0:
k = row.split(' => ')[0].strip().replace("'", '')
v = row.split(' => ')[1].strip().replace("'", '').replace(',', '')
output[k] = v
debug('Perl magic vars: %s' % output)
debug('Metric associations: %s' % dict((k, output[v]) for (k, v) in metrics.iteritems()))
# Handle metrics
results = []
for metric in selected_metrics:
stats = rds.get_metric(metric)
if metric == 'FreeableMemory':
info = rds.get_info()
try:
memory = db_classes[info.instance_class] * 1024 ** 3
except IndexError:
print 'Unknown DB instance class "%s"' % info.instance_class
sys.exit(1)
results.append('%s:%.0f' % (output['used_memory'], memory - stats))
results.append('%s:%.0f' % (output['total_memory'], memory))
elif metric == 'FreeStorageSpace':
info = rds.get_info()
storage = float(info.allocated_storage) * 1024 ** 3
results.append('%s:%.0f' % (output['used_space'], storage - stats))
results.append('%s:%.0f' % (output['total_space'], storage))
else:
short_var = output.get(metrics[metric])
if not short_var:
print 'Chosen metric does not have a correspondent entry in perl magic vars'
sys.exit(1)
results.append('%s:%s' % (short_var, stats))
print ' '.join(results)
def main():
"""Main function"""
global options
# DB instance classes as listed on
# http://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/Concepts.DBInstanceClass.html
db_classes = {
'db.t1.micro': 0.615,
'db.m1.small': 1.7,
'db.m1.medium': 3.75,
'db.m1.large': 7.5,
'db.m1.xlarge': 15,
'db.m4.large': 8,
'db.m4.xlarge': 16,
'db.m4.2xlarge': 32,
'db.m4.4xlarge': 64,
'db.m4.10xlarge': 160,
'db.r3.large': 15,
'db.r3.xlarge': 30.5,
'db.r3.2xlarge': 61,
'db.r3.4xlarge': 122,
'db.r3.8xlarge': 244,
'db.t2.micro': 1,
'db.t2.small': 2,
'db.t2.medium': 4,
'db.t2.large': 8,
'db.m3.medium': 3.75,
'db.m3.large': 7.5,
'db.m3.xlarge': 15,
'db.m3.2xlarge': 30,
'db.m2.xlarge': 17.1,
'db.m2.2xlarge': 34.2,
'db.m2.4xlarge': 68.4,
'db.cr1.8xlarge': 244,
}
# RDS metrics http://docs.aws.amazon.com/AmazonCloudWatch/latest/DeveloperGuide/rds-metricscollected.html
metrics = {
'BinLogDiskUsage':
'binlog_disk_usage', # The amount of disk space occupied by binary logs on the master. Units: Bytes
'CPUUtilization':
'utilization', # The percentage of CPU utilization. Units: Percent
'DatabaseConnections':
'connections', # The number of database connections in use. Units: Count
'DiskQueueDepth':
'disk_queue_depth', # The number of outstanding IOs (read/write requests) waiting to access the disk. Units: Count
'ReplicaLag':
'replica_lag', # The amount of time a Read Replica DB Instance lags behind the source DB Instance. Units: Seconds
'SwapUsage':
'swap_usage', # The amount of swap space used on the DB Instance. Units: Bytes
'FreeableMemory':
'used_memory', # The amount of available random access memory. Units: Bytes
'FreeStorageSpace':
'used_space', # The amount of available storage space. Units: Bytes
'ReadIOPS':
'read_iops', # The average number of disk I/O operations per second. Units: Count/Second
'WriteIOPS':
'write_iops', # The average number of disk I/O operations per second. Units: Count/Second
'ReadLatency':
'read_latency', # The average amount of time taken per disk I/O operation. Units: Seconds
'WriteLatency':
'write_latency', # The average amount of time taken per disk I/O operation. Units: Seconds
'ReadThroughput':
'read_throughput', # The average number of bytes read from disk per second. Units: Bytes/Second
'WriteThroughput':
'write_throughput', # The average number of bytes written to disk per second. Units: Bytes/Second
}
# Parse options
parser = optparse.OptionParser()
parser.add_option('-l',
'--list',
help='list DB instances',
action='store_true',
default=False,
dest='db_list')
parser.add_option(
'-n',
'--profile',
default=None,
help=
'AWS profile from ~/.boto or /etc/boto.cfg. Default: None, fallbacks to "[Credentials]".'
)
parser.add_option(
'-r',
'--region',
default='us-east-1',
help=
'AWS region. Default: us-east-1. If set to "all", we try to detect the instance region '
'across all of them, note this will be slower than if you specify the region explicitly.'
)
parser.add_option('-i', '--ident', help='DB instance identifier')
parser.add_option(
'-p',
'--print',
help='print status and other details for a given DB instance',
action='store_true',
default=False,
dest='printinfo')
parser.add_option('-m',
'--metric',
help='metrics to retrive separated by comma: [%s]' %
', '.join(metrics.keys()))
parser.add_option('-d',
'--debug',
help='enable debugging',
action='store_true',
default=False)
options, _ = parser.parse_args()
# Strip a prefix _ which is sent by Cacti, so an empty argument is interpreted correctly.
# Than set defaults if argument is supposed to be empty.
options.region = options.region.lstrip('_')
options.profile = options.profile.lstrip('_')
if not options.region:
options.region = 'us-east-1'
if not options.profile:
options.profile = None
if options.debug:
boto.set_stream_logger('boto')
rds = RDS(region=options.region,
profile=options.profile,
identifier=options.ident)
# Check args
if len(sys.argv) == 1:
parser.print_help()
sys.exit()
elif options.db_list:
info = rds.get_list()
print 'List of all DB instances in %s region(s):' % (options.region, )
pprint.pprint(info)
sys.exit()
elif not options.ident:
parser.print_help()
parser.error('DB identifier is not set.')
elif options.printinfo:
info = rds.get_info()
pprint.pprint(vars(info))
sys.exit()
elif not options.metric:
parser.print_help()
parser.error('Metric is not set.')
selected_metrics = options.metric.split(',')
for metric in selected_metrics:
if metric not in metrics.keys():
parser.print_help()
parser.error('Invalid metric.')
# Do not remove the empty lines in the start and end of this docstring
perl_magic_vars = """
# Define the variables to output. I use shortened variable names so maybe
# it'll all fit in 1024 bytes for Cactid and Spine's benefit. Strings must
# have some non-hex characters (non a-f0-9) to avoid a Cacti bug. This list
# must come right after the word MAGIC_VARS_DEFINITIONS. The Perl script
# parses it and uses it as a Perl variable.
$keys = array(
'binlog_disk_usage' => 'gg',
'utilization' => 'gh',
'connections' => 'gi',
'disk_queue_depth' => 'gj',
'replica_lag' => 'gk',
'swap_usage' => 'gl',
'used_memory' => 'gm',
'total_memory' => 'gn',
'used_space' => 'go',
'total_space' => 'gp',
'read_iops' => 'gq',
'write_iops' => 'gr',
'read_latency' => 'gs',
'write_latency' => 'gt',
'read_throughput' => 'gu',
'write_throughput' => 'gv',
);
"""
output = dict()
for row in perl_magic_vars.split('\n'):
if row.find('=>') >= 0:
k = row.split(' => ')[0].strip().replace("'", '')
v = row.split(' => ')[1].strip().replace("'", '').replace(',', '')
output[k] = v
debug('Perl magic vars: %s' % output)
debug('Metric associations: %s' % dict(
(k, output[v]) for (k, v) in metrics.iteritems()))
# Handle metrics
results = []
for metric in selected_metrics:
stats = rds.get_metric(metric)
if metric == 'FreeableMemory':
info = rds.get_info()
try:
memory = db_classes[info.instance_class] * 1024**3
except IndexError:
print 'Unknown DB instance class "%s"' % info.instance_class
sys.exit(1)
results.append('%s:%.0f' % (output['used_memory'], memory - stats))
results.append('%s:%.0f' % (output['total_memory'], memory))
elif metric == 'FreeStorageSpace':
info = rds.get_info()
storage = float(info.allocated_storage) * 1024**3
results.append('%s:%.0f' % (output['used_space'], storage - stats))
results.append('%s:%.0f' % (output['total_space'], storage))
else:
short_var = output.get(metrics[metric])
if not short_var:
print 'Chosen metric does not have a correspondent entry in perl magic vars'
sys.exit(1)
results.append('%s:%s' % (short_var, stats))
print ' '.join(results)