Awesome

Nagios Plugins by AppFirst

Nagios Plugins for popular services/servers hosted by different Cloud Services such as EngineYard, Heroku. Seamless integration with AppFirst Collector. Now with support to streaming Statsd Data with Statsd_clients over AFCollectorAPI or UDP protocol.

This README will cover the following topics:

• Nagios Plugins
  • check_postgresql (and check_postgresql_92)
  • check_mysql
  • check_mongodb
  • check_memcached
  • check_redis
  • check_resque
  • check_passenger
• Data Module and Base Classes
• StastD Wrapping
• AppFirst Integration

For further question, please email clark@appfirst.com

##Nagios Plugins

To fetch the data for those plugins, there are mainly two kinds of way depends on what the service provides:

• BATCH to fetch multiple metrics in one query (such as return a dictionary or even a tree)
• DEDICATED to fetch individual metrics with dedicated queries.

At the moment, there are 2 kinds of statistical value in general.

• STATUS value, the value at the moment of checking.
• DELTA value, the changes between now and the last time of checking. In most cases, these values are not provided by the service/servers. The services are more likely to provide a value since the server is started. Those values, such as total connections, total operations or total bytes received are incremental values, so it's impossible to set a threshold for warning and critical (as eventually it will pass over any value given). Thus we'll have to calculate the delta by comparing the current value with previously checked value. And we'll also have to store the current one for future comparison. By defining a rootdir and a filename with -d ROOTDIR -f FILENAME, the data will be store on disk. By default, rootdir is /tmp and the filenames looks like pd@cmd2get_data. For the first time running the script or if the file is not accessible (removed, rootdir changed, etc), the output value will not be delta value since there is no previous value referrable.

Values are either one of:

• SINGLE Most data sources has one value for each metrics.
• MULTIPLE Some data sources have provided more than one statistical value, i.e. sub-performance value, such as values of individual databases or of different types. We usually trail those after the total value.

In order to use any of the plugin, make sure you have nagios.py and statsd.py in PYTHONPATH or in the same folder with the check_*.py plugins you'll use.

###PostgreSQL

Check status for various metrics of PostgresSQL, with commandline access:

python check_postgresql.py <options>

the options are:

usage: check_postgresql.py [-h] [-w WARN] [-c CRIT] -t
                           {DATABASE_SIZE, TUPLES_UPDATED, LOCKS_ROW, LOCKS_ACCESS, TUPLES_READ, LOCKS_SHARE, CONNECTIONS_WAITING, CONNECTIONS_IDLE, CONNECTIONS_ACTIVE, LOCKS_EXCLUSIVE, TUPLES_INSERTED, TUPLES_DELETED}
                           [-d ROOTDIR] [-f FILENAME] [-u USER] [-p PORT]
                           [-z APPNAME]                               

Note: If you are running PostgreSQL 9.2.x, you should use the check_postgresql_92.py script instead of check_postgresql.py. The options and arguments are exactly the same for both scripts. There is a separate script for 9.2 because PostgreSQL 9.2 introduced some changes to the statistics tables which require slightly different queries.

TUPLES_UPDATED

dedicated | multiple | delta

This will check tuples updated in every databases of postgres since last check.
Sample output:

TUPLES_UPDATED OK: 0 tuples updated | total=0 template1=0 template0=0 postgres=0 appfirst=0

TUPLES_READ

dedicated | multiple | delta

This will check tuples fetched in every databases of postgres since last check.
Sample output:

TUPLES_READ OK: 505 tuples fetched | total=505 template1=0 template0=0 postgres=505 appfirst=0

TUPLES_INSERTED

dedicated | multiple | delta

This will check tuples inserted in every databases of postgres since last check.
Sample output:

TUPLES_INSERTED OK: 0 tuples inserted | total=0 template1=0 template0=0 postgres=0 appfirst=0

TUPLES_DELETED

dedicated | multiple | delta

This will check tuples deleted in every databases of postgres since last check.
Sample output:

TUPLES_DELETED OK: 0 tuples deleted | total=0 template1=0 template0=0 postgres=0 appfirst=0

DATABASE_SIZE

dedicated | multiple | status

This will check size (in MB) of every databases of postgres since last check.
Sample output:

DATABASE_SIZE OK: total dbsize: 22MB | total=22MB template1=5 template0=5 postgres=5 appfirst=5

CONNECTIONS_WAITING

dedicated | single | status

This will check how many connections are currently waiting.
Sample output:

CONNECTIONS_WAITING OK: 0 waiting conns | waiting=0;20;40

CONNECTIONS_IDLE

dedicated | single | status

This will check how many connections are currently idle.
Sample output:

CONNECTIONS_IDLE OK: 0 idle conns | idle=0;20;40

CONNECTIONS_ACTIVE

dedicated | single | status

This will check how many connections are currently active.
Sample output:

CONNECTIONS_ACTIVE OK: 1 active conns | active=1;100;500

###MySQL

Check status for various metrics of MySQL, with commandline access:

python check_mysql.py <options>

the options are:

usage: check_mysql.py [-h] [-w WARN] [-c CRIT] -t
                      {TRANSACTIONS, SLOW_QUERIES, ROW_OPERATIONS, CONNECTIONS, REPLICATION, TOTAL_BYTES, NETWORK_TRAFFIC, QUERIES_PER_SECOND, SELECTS}
                      [-d ROOTDIR] [-f FILENAME] [-u USER] [-s PASSWORD]
                      [-H HOST] [-p PORT] [-z APPNAME]

TRANSACTIONS

batch | multiple | delta

This will check how many transtions

• commit transoactions
• rollback transactions

respectively.
Sample output:

TRANSACTIONS OK: 0 transactions | total=0 commit=0 rollback=0

TOTAL_BYTES

batch | multiple | delta

This will check how many

• bytes received
• bytes sent

respectively.
Sample output:

TOTAL_BYTES OK: 0.0557985305786MB in total | total=0.0557985305786MB bytes_received=0.000700950622559MB bytes_sent=0.0550975799561MB

SLOW_QUERIES

batch | single | delta

This will check how many slow queries.
Sample output:

SLOW_QUERIES OK: 0 slow queries | total=0

SELECTS

batch | multiple | delta

This will check selects including

• select full join
• select full range join
• select range
• select range check
• select scan

Sample output:

SELECTS OK: 7 select | total=7 select_full_join=0 select_full_range_join=0 select_range=0 select_range_check=0 select_scan=7

ROW_OPERATIONS

batch | multiple | delta

This will check row operations including

• rows deleted
• rows inserted
• rows updated
• rows read

Sample output:

ROW_OPERATIONS OK: 0 row operations | total=0 rows_deleted=0 rows_inserted=0 rows_updated=0 rows_read=0

REPLICATION

This TYPE is incomplete.
Sample output:

REPLICATION UNKNOWN: mysterious status

QUERIES_PER_SECOND

batch | single | delta

This is the rate of queries (per sec) since last check.
Sample output:

QUERIES_PER_SECOND OK: 0.0448717948718 queries per second | total=0.0448717948718;100;300

CONNECTIONS

batch | single | delta

This checks the connections established since last check. Sample output:

CONNECTIONS OK: 7 new connections | conns=7;100;300

###MongoDB

Check status for various metrics of MongoDB, with commandline access:

python check_mongodb.py <options>

the options are:

usage: check_mongodb.py [-h] [-w WARN] [-c CRIT] -t
                        {MEMORY_USED, INSERT, HITS, MISS_PERCENTAGE, LOCKED_PERCENTAGE, UPDATE, ACCESSES, CONNECTIONS, MISSES, COMMAND, QUERY, RESETS, DELETE}
                        [-d ROOTDIR] [-f FILENAME] [-u USER] [-s PASSWORD]
                        [-H HOST] [-p PORT] [-z APPNAME]

CONNECTIONS

dedicated | single | status

Sample output:

CONNECTIONS OK: 1 new connections | conns=1;100;200

MEMORY_USED

dedicated | single | status

Sample output:

MEMORY_USED OK: 12MB resident size | res=12MB

INSERT

dedicated | single | delta

Sample output:

INSERT OK: 0 inserts | inserts=0

UPDATE

dedicated | single | delta

Sample output:

UPDATE OK: 0 updates | updates=0

COMMAND

dedicated | single | delta

Sample output:

COMMAND OK: 15168 commands | commands=15168

QUERY

dedicated | single | delta

Sample output:

QUERY OK: 1 queries | queries=1

DELETE

dedicated | single | delta

Sample output:

DELETE OK: 0 deletes | deletes=0

LOCKED_PERCENTAGE

dedicated | single | delta

Sample output:

DELETE OK: 0 locked | ratio=0%

MISS_PERCENTAGE

dedicated | single | delta

Sample output:

DELETE OK: 0 missed | ratio=0%

RESETS

dedicated | single | delta

Sample output:

RESETS OK: 0 resets | resets=0

HITS

dedicated | single | delta

Sample output:

HITS OK: 0 hits | hits=0

MISSES

dedicated | single | delta

Sample output:

MISSES OK: 0 misses | misses=0

ACCESSES

dedicated | single | delta

Sample output:

ACCESSES OK: 0 accesses | accesses=0

###MemcacheD

Check status for various metrics of MemcacheD, with commandline access:

python check_memcached.py <options>

the options are:

usage: check_memcached.py [-h] [-w WARN] [-c CRIT] -t
                          {TOTAL_ITEMS, BYTES_ALLOCATED, OPERATIONS_SET_REQUESTS, BYTES_WRITTEN, BYTES_READ, CURRENT_CONNECTIONS, OPERATIONS_GET_REQUESTS}
                          [-d ROOTDIR] [-f FILENAME] [-H HOST] [-p PORT]
                          [-z APPNAME]

TOTAL_ITEMS

batch | single | status

Sample output:

TOTAL_ITEMS OK: 0 total items | items=0

CURRENT_CONNECTIONS

batch | single | status

Sample output:

CURRENT_CONNECTIONS OK: 5 current connections | connections=5

OPERATIONS_SET_REQUESTS

batch | single | delta

Sample output:

OPERATIONS_SET_REQUESTS OK: 0 set requests | set_requests=0

OPERATIONS_GET_REQUESTS

batch | single | delta

Sample output:

OPERATIONS_GET_REQUESTS OK: 0 get resquests | get_requests=0

BYTES_WRITTEN

batch | single | delta

Sample output:

BYTES_WRITTEN OK: 6171 bytes written | bytes_written=6171

BYTES_READ

batch | single | delta

Sample output:

BYTES_READ OK: 66 bytes read | bytes_read=66

BYTES_ALLOCATED

batch | single | status

Sample output:

BYTES_ALLOCATED OK: 0 bytes allocated | bytes_allocated=0

###Redis

Check status for various metrics of Redis, with commandline access:

python check_redis.py <options>

the options are:

usage: check_redis.py [-h] [-w WARN] [-c CRIT] -t
                      {MEMORY_USED, CURRENT_OPERATIONS, CHANGES_SINCE_LAST_SAVE, READ_WRITE_RATIO, AVERAGE_OPERATIONS_RATE, COMMAND_FREQUENCY, CURRENT_CHANGES, TOTAL_KEYS}
                      [-d ROOTDIR] [-f FILENAME] [-u USER] [-s PASSWORD]
                      [-H HOST] [-p PORT] [-n DATABASE] [-z APPNAME]

TOTAL_KEYS

dedicated | single | status

Sample output:

TOTAL_KEYS OK: 8 total keys | total_keys=8;1000000;2000000

MEMORY_USED

batch | single | status

Sample output:

MEMORY_USED OK: 0MB used_memory | used_memory=0MB;2;4

AVERAGE_OPERATIONS_RATE

batch | single | status

Sample output:

AVERAGE_OPERATIONS_RATE OK: 0 commands per second | average_rate=0;20000;40000

CURRENT_OPERATIONS

batch | single | delta

Sample output:

CURRENT_OPERATIONS OK: 10 commands | current_commands=10;20000;40000

CURRENT_CHANGES

batch | single | delta

Sample output:

CURRENT_CHANGES OK: 0 changes | changes=0

CHANGES_SINCE_LAST_SAVE

batch | single | status

Sample output:

CHANGES_SINCE_LAST_SAVE OK: 0 changes since last save | changes=0;100;300

###Resque

Check status for various metrics of Resque, with commandline access:

python check_resque.py <options>

the options are:

usage: check_resque.py [-h] [-w WARN] [-c CRIT] -t
                       {QUEUE_LENGTH,JOB_PROCESSED} [-d ROOTDIR] [-f FILENAME]
                       [-u USER] [-s PASSWORD] [-H HOST] [-p PORT]
                       [-n DATABASE] [-z APPNAME]

QUEUE_LENGTH

dedicated | multiple | status

Sample output:

QUEUE_LENGTH OK: 0 jobs in queues | total=0 default=0 critical=0 failing=0

JOB_PROCESSED

dedicated | single | delta

Sample output:

JOB_PROCESSED OK: 0 jobs in processed | total=133

###Passenger

Check status for various metrics of Passenger, with commandline access:

python check_passenger.py <options>

the options are:

usage: check_passenger.py [-h] [-w WARN] [-c CRIT] -t
                          {ACTIVE_PROCESSES,MAX_PROCESSES,RUNNING_PROCESSES}
                          [-d ROOTDIR] [-f FILENAME] [-p PID] [-z APPNAME]

RUNNING_PROCESSES

batch | single | status

Sample output:

RUNNING_PROCESSES UNKNOWN: ERROR: Phusion Passenger doesn't seem to be running.

MAX_PROCESSES

batch | single | status

Sample output:

MAX_PROCESSES UNKNOWN: ERROR: Phusion Passenger doesn't seem to be running.

ACTIVE_PROCESSES

batch | single | status

Sample output:

ACTIVE_PROCESSES UNKNOWN: ERROR: Phusion Passenger doesn't seem to be running.

##Data Module and Base Classes

nagios.py holds all the datatype class and base class for plugins.

@plugin.command("BYTES_WRITTEN")

##StatsD Wrapping

statsd.py provides plugins an easy way to send performance value as statsd message. This module requires those files in statsd_clients/python in PYTHON_PATH.

There are currently three statsd wrapper: statsd.gauge, statsd.counter, statsd.timer. The usage is quite intuitive, just add it to any method that returns a nagios.Result with a performance value. The wrapper will intercept the return value of the method, make the bucket name from the BUCKET_PATTERN (by default it's "sys.app.%(appname)s.%(type)s"). You can

statsd.set_bucket_pattern(pattern)

to your own pattern. Available items from Result are: appname, type, status_code, status, message, value (the first of performance values). By default this wrapper will send statsd to AppFirst Collector through AFTransport, but you can pass in your own transport by

statsd.set_transport(transport)

Please reference details to Statsd_clients.

For plugins that has STATUS value, gauge is most likely to be suitable, those usually include memory usage, rate, active connection and total keys in databses.
For plugins that has DELTA value, counter is most likely to be suitable, those usually include bytes transfered, operations, errors, newly established connections.

Here's an example of counter:

@nagios.CommandBasedPlugin.command("BYTES_WRITTEN")
@statsd.counter
def get_bytes_written(self, request):
    return nagios.Result(type, status_code)

Note that the statsd descriptor should be applied after the nagios.CommandBasedPlugin.command descriptor in order to be executed if you are writing an CommandBasedPlugin.

##AppFirst Integration

##Appendix:

Here's TABLE OF COMPARISON OF METRICS:

           | TUPLES_READ             | dedicated   | multiple   | delta       | counter
           | TUPLES_INSERTED         | dedicated   | multiple   | delta       | counter
           | TUPLES_DELETED          | dedicated   | multiple   | delta       | counter
           | LOCKS_ACCESS            | dedicated   | multiple   | status      | gauge
           | LOCKS_ROW               | dedicated   | multiple   | status      | gauge
           | LOCKS_SHARE             | dedicated   | multiple   | status      | gauge
           | LOCKS_EXCLUSIVE         | dedicated   | multiple   | status      | gauge
           | DATABASE_SIZE           | dedicated   | multiple   | status      | gauge
           | CONNECTIONS_WAITING     | dedicated   | single     | status      | gauge
           | CONNECTIONS_IDLE        | dedicated   | single     | status      | gauge
           | CONNECTIONS_ACTIVE      | dedicated   | single     | status      | gauge

MySQL | TRANSACTIONS | batch | multiple | delta | counter | TOTAL_BYTES | batch | multiple | delta | counter | SLOW_QUERIES | batch | single | delta | counter | SELECTS | batch | multiple | delta | counter | ROW_OPERATIONS | batch | multiple | delta | counter | QUERIES_PER_SECOND | batch | single | delta | gauge | CONNECTIONS | batch | single | delta | counter MongoDB | CONNECTIONS | dedicated | single | status | gauge | MEMORY_USED | dedicated | single | status | gauge | INSERT | dedicated | single | status | counter | UPDATE | dedicated | single | status | counter | COMMAND | dedicated | single | status | counter | QUERY | dedicated | single | status | counter | DELETE | dedicated | single | status | counter | LOCKED_PERCENTAGE | batch | single | status | gauge | MISS_PERCENTAGE | dedicated | single | status | gauge | HITS | dedicated | single | status | counter | MISSES | dedicated | single | status | counter | RESETS | dedicated | single | status | counter | ACCESSES | dedicated | single | status | counter Memcached | TOTAL_ITEMS | batch | single | status | gauge | CURRENT_CONNECTIONS | batch | single | status | gauge | OPERATIONS_SET_REQUESTS | batch | single | delta | counter | OPERATIONS_GET_REQUESTS | batch | single | delta | counter | BYTES_WRITTEN | batch | single | delta | counter | BYTES_READ | batch | single | delta | counter | BYTES_ALLOCATED | batch | single | status | gauge Redis | TOTAL_ITEMS | dedicated | single | status | gauge | MEMORY_USED | batch | single | status | gauge | AVERAGE_OPERATIONS_RATE | batch | single | status | gauge | CURRENT_OPERATIONS | batch | single | status | gauge | CURRENT_CHANGES | batch | single | delta | counter | CHANGES_SINCE_LAST_SAVE | batch | single | status | gauge Resque | QUEUE_LENGTH | dedicated | multiple | status | gauge | JOB_PROCESSED | batch | single | delta | counter Passenger | RUNNING_PROCESSES | batch | single | status | gauge | MAX_PROCESSES | batch | single | status | gauge | ACTIVE_PROCESSES | batch | single | status | gauge