pmp-check-aws-rds.py

#!/usr/bin/env python
"""Nagios plugin for Amazon RDS monitoring.

https://github.com/percona/percona-monitoring-plugins

This program is part of Percona Monitoring Plugins
License: GPL License (see COPYING)

Author Roman Vynar
Copyright 2014-2015 Percona LLC and/or its affiliates
"""

import datetime
import optparse
import pprint
import sys

import boto
import boto.rds
import boto.ec2.cloudwatch

# Nagios status codes
OK = 0
WARNING = 1
CRITICAL = 2
UNKNOWN = 3


class RDS(object):

    """RDS connection class"""

    def __init__(self, region, profile=None, identifier=None):
        """Get RDS instance details"""
        self.region = region
        self.profile = profile
        self.identifier = identifier

        if self.region == 'all':
            self.regions_list = [reg.name for reg in boto.rds.regions()]
        else:
            self.regions_list = [self.region]

        self.info = None
        if self.identifier:
            for reg in self.regions_list:
                try:
                    rds = boto.rds.connect_to_region(reg, profile_name=self.profile)
                    self.info = rds.get_all_dbinstances(self.identifier)
                except (boto.provider.ProfileNotFoundError, boto.exception.BotoServerError) as msg:
                    debug(msg)
                else:
                    # Exit on the first region and identifier match
                    self.region = reg
                    break

    def get_info(self):
        """Get RDS instance info"""
        if self.info:
            return self.info[0]
        else:
            return None

    def get_list(self):
        """Get list of available instances by region(s)"""
        result = dict()
        for reg in self.regions_list:
            try:
                rds = boto.rds.connect_to_region(reg, profile_name=self.profile)
                result[reg] = rds.get_all_dbinstances()
            except (boto.provider.ProfileNotFoundError, boto.exception.BotoServerError) as msg:
                debug(msg)

        return result

    def get_metric(self, metric, start_time, end_time, step):
        """Get RDS metric from CloudWatch"""
        cw_conn = boto.ec2.cloudwatch.connect_to_region(self.region, profile_name=self.profile)
        result = cw_conn.get_metric_statistics(
            step,
            start_time,
            end_time,
            metric,
            'AWS/RDS',
            'Average',
            dimensions={'DBInstanceIdentifier': [self.identifier]}
        )
        if result:
            if len(result) > 1:
                # Get the last point
                result = sorted(result, key=lambda k: k['Timestamp'])
                result.reverse()

            result = float('%.2f' % result[0]['Average'])

        return result


def debug(val):
    """Debugging output"""
    global options
    if options.debug:
        print 'DEBUG: %s' % val


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.m4.16xlarge': 256,
        'db.m5.large': 8,
        'db.m5.xlarge': 16,
        'db.m5.2xlarge': 32,
        'db.m5.4xlarge': 64,
        'db.m5.12xlarge': 192,
        'db.m5.24xlarge': 384,
        'db.r3.large': 15,
        'db.r3.xlarge': 30.5,
        'db.r3.2xlarge': 61,
        'db.r3.4xlarge': 122,
        'db.r3.8xlarge': 244,
        'db.r4.large': 15.25,
        'db.r4.xlarge': 30.5,
        'db.r4.2xlarge': 61,
        'db.r4.4xlarge': 122,
        'db.r4.8xlarge': 244,
        'db.r4.16xlarge': 488,
        'db.r5.large': 16,
        'db.r5.xlarge': 32,
        'db.r5.2xlarge': 64,
        'db.r5.4xlarge': 128,
        'db.r5.8xlarge': 256,
        'db.r5.12xlarge': 384,
        'db.r5.16xlarge': 512,
        'db.r5.24xlarge': 768,
        'db.t2.micro': 1,
        'db.t2.small': 2,
        'db.t2.medium': 4,
        'db.t2.large': 8,
        'db.t2.xlarge': 16,
        'db.t2.2xlarge': 32,
        'db.t3.micro': 1,
        'db.t3.small': 2,
        'db.t3.medium': 4,
        'db.t3.large': 8,
        'db.t3.xlarge': 16,
        'db.t3.2xlarge': 32,
        '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,
        'db.x1.16xlarge': 976,
        'db.x1.32xlarge': 1952,
        'db.x1e.xlarge': 122,
        'db.x1e.2xlarge': 244,
        'db.x1e.4xlarge': 488,
        'db.x1e.8xlarge': 976,
        'db.x1e.16xlarge': 1952,
        'db.x1e.32xlarge': 3904,
    }

    # 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('-f', '--forceunknown', help='force alerts on unknown status. This prevents issues related to '
                      'AWS Cloudwatch throttling limits Default: False',
                      action='store_true', default=False)
    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)
    elif status == UNKNOWN and not options.forceunknown:
        print '%s %s | null' % ('OK', note)
        sys.exit(0)
    else:
        print '%s %s' % (short_status[status], note)

    sys.exit(status)


if __name__ == '__main__':
    main()

# ############################################################################
# Documentation
# ############################################################################
"""
=pod

=head1 NAME

pmp-check-aws-rds.py - Check Amazon RDS metrics.

=head1 SYNOPSIS

  Usage: pmp-check-aws-rds.py [options]

  Options:
    -h, --help            show this help message and exit
    -l, --list            list of all DB instances
    -n PROFILE, --profile-name=PROFILE
                          AWS profile from ~/.boto or /etc/boto.cfg. Default:
                          None, fallbacks to "[Credentials]".
    -r REGION, --region=REGION
                          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 you specify the region.
    -i IDENT, --ident=IDENT
                          DB instance identifier
    -p, --print           print status and other details for a given DB instance
    -m METRIC, --metric=METRIC
                          metric to check: [status, load, storage, memory]
    -w WARN, --warn=WARN  warning threshold
    -c CRIT, --crit=CRIT  critical threshold
    -u UNIT, --unit=UNIT  unit of thresholds for "storage" and "memory" metrics:
                          [percent, GB]. Default: percent
    -t TIME, --time=TIME  time period in minutes to query. Default: 5
    -a AVG, --avg=AVG     time average in minutes to request. Default: 1
    -f, --forceunknown    force alerts on unknown status. This prevents issues
                          related to AWS Cloudwatch throttling limits Default:
                          False
    -d, --debug           enable debug output

=head1 REQUIREMENTS

This plugin is written on Python and utilizes the module C<boto> (Python interface
to Amazon Web Services) to get various RDS metrics from CloudWatch and compare
them against the thresholds.

* Install the package: C<yum install python-boto> or C<apt-get install python-boto>
* Create a config /etc/boto.cfg or ~nagios/.boto with your AWS API credentials.
  See http://code.google.com/p/boto/wiki/BotoConfig

This plugin that is supposed to be run by Nagios, i.e. under ``nagios`` user,
should have permissions to read the config /etc/boto.cfg or ~nagios/.boto.

Example:

  [root@centos6 ~]# cat /etc/boto.cfg
  [Credentials]
  aws_access_key_id = THISISATESTKEY
  aws_secret_access_key = thisisatestawssecretaccesskey

If you do not use this config with other tools such as our Cacti script,
you can secure this file the following way:

  [root@centos6 ~]# chown nagios /etc/boto.cfg
  [root@centos6 ~]# chmod 600 /etc/boto.cfg

=head1 DESCRIPTION

The plugin provides 4 checks and some options to list and print RDS details:

* RDS Status
* RDS Load Average
* RDS Free Storage
* RDS Free Memory

To get the list of all RDS instances under AWS account:

  # ./pmp-check-aws-rds.py -l

To get the detailed status of RDS instance identified as C<blackbox>:

  # ./pmp-check-aws-rds.py -i blackbox -p

Nagios check for the overall status. Useful if you want to set the rest
of the checks dependent from this one:

  # ./pmp-check-aws-rds.py -i blackbox -m status
  OK mysql 5.1.63. Status: available

Nagios check for CPU utilization, specify thresholds as percentage of
1-min., 5-min., 15-min. average accordingly:

  # ./pmp-check-aws-rds.py -i blackbox -m load -w 90,85,80 -c 98,95,90
  OK Load average: 18.36%, 18.51%, 15.95% | load1=18.36;90.0;98.0;0;100 load5=18.51;85.0;95.0;0;100 load15=15.95;80.0;90.0;0;100

Nagios check for the free memory, specify thresholds as percentage:

  # ./pmp-check-aws-rds.py -i blackbox -m memory -w 5 -c 2
  OK Free memory: 5.90 GB (9%) of 68 GB | free_memory=8.68;5.0;2.0;0;100
  # ./pmp-check-aws-rds.py -i blackbox -m memory -u GB -w 4 -c 2
  OK Free memory: 5.90 GB (9%) of 68 GB | free_memory=5.9;4.0;2.0;0;68

Nagios check for the free storage space, specify thresholds as percentage or GB:

  # ./pmp-check-aws-rds.py -i blackbox -m storage -w 10 -c 5
  OK Free storage: 162.55 GB (33%) of 500.0 GB | free_storage=32.51;10.0;5.0;0;100
  # ./pmp-check-aws-rds.py -i blackbox -m storage -u GB -w 10 -c 5
  OK Free storage: 162.55 GB (33%) of 500.0 GB | free_storage=162.55;10.0;5.0;0;500.0

By default, the region is set to ``us-east-1``. You can re-define it globally in boto config or
specify with -r option. The following command will list all instances across all regions under your AWS account:

  # ./pmp-check-aws-rds.py -r all -l

The following command will show the status for the first instance identified as ``blackbox`` in all regions:

  # ./pmp-check-aws-rds.py -r all -i blackbox -p

Remember, scanning regions are slower operation than specifying it explicitly.

=head1 CONFIGURATION

Here is the excerpt of potential Nagios config:

  define host{
        use                             mysql-host
        host_name                       blackbox
        alias                           blackbox
        address                         blackbox.abcdefgh.us-east-1.rds.amazonaws.com
        }

  define servicedependency{
        host_name                       blackbox
        service_description             RDS Status
        dependent_service_description   RDS Load Average, RDS Free Storage, RDS Free Memory
        execution_failure_criteria      w,c,u,p
        notification_failure_criteria   w,c,u,p
        }

  define service{
        use                             active-service
        host_name                       blackbox
        service_description             RDS Status
        check_command                   check_rds!status!0!0
        }

  define service{
        use                             active-service
        host_name                       blackbox
        service_description             RDS Load Average
        check_command                   check_rds!load!90,85,80!98,95,90
        }

  define service{
        use                             active-service
        host_name                       blackbox
        service_description             RDS Free Storage
        check_command                   check_rds!storage!10!5
        }

  define service{
        use                             active-service
        host_name                       blackbox
        service_description             RDS Free Memory
        check_command                   check_rds!memory!5!2
        }

  define command{
        command_name    check_rds
        command_line    $USER1$/pmp-check-aws-rds.py -i $HOSTALIAS$ -m $ARG1$ -w $ARG2$ -c $ARG3$
        }

=head1 COPYRIGHT, LICENSE, AND WARRANTY

This program is copyright 2014 Percona LLC and/or its affiliates.
Feedback and improvements are welcome.

THIS PROGRAM IS PROVIDED "AS IS" AND WITHOUT ANY EXPRESS OR IMPLIED
WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.

This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
Foundation, version 2.  You should have received a copy of the GNU General
Public License along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA.

=head1 VERSION

This program is part of Percona Monitoring Plugins pmp-check-aws-rds.py 1.1.6

=cut

"""