Beaver

python daemon that munches on logs and sends their contents to logstash

Requirements

• Python 2.6+
• Optional zeromq support: install libzmq (brew install zmq or apt-get install libzmq-dev) and pyzmq (pip install pyzmq==2.1.11)

Installation

Using PIP:

From Github:

pip install git+git://github.com/josegonzalez/beaver.git#egg=beaver

From PyPI:

pip install beaver==22

Usage

usage:

beaver [-h] [-c CONFIG] [-d] [-D] [-f FILES [FILES ...]]
       [-F {json,msgpack,string}] [-H HOSTNAME] [-m {bind,connect}]
       [-o OUTPUT] [-p PATH] [-P PID]
       [-t {rabbitmq,redis,stdout,zmq,udp}] [-v] [--fqdn]

optional arguments:

-h, --help            show this help message and exit
-c CONFIG, --configfile CONFIG
                      ini config file path
-d, --debug           enable debug mode
-D, --daemonize       daemonize in the background
-f FILES [FILES ...], --files FILES [FILES ...]
                      space-separated filelist to watch, can include globs
                      (*.log). Overrides --path argument
-F {json,msgpack,string}, --format {json,msgpack,string}
                      format to use when sending to transport
-H HOSTNAME, --hostname HOSTNAME
                      manual hostname override for source_host
-m {bind,connect}, --mode {bind,connect}
                      bind or connect mode
-o OUTPUT, --output OUTPUT
                      file to pipe output to (in addition to stdout)
-p PATH, --path PATH  path to log files
-P PID, --pid PID     path to pid file
-t {rabbitmq,redis,stdout,zmq,udp}, --transport {rabbitmq,redis,stdout,zmq,udp}
                      log transport method
-v, --version         output version and quit
--fqdn                use the machine's FQDN for source_host

Background

Beaver provides an lightweight method for shipping local log files to Logstash. It does this using either redis, stdin, zeromq as the transport. This means you'll need a redis, stdin, zeromq input somewhere down the road to get the events.

Events are sent in logstash's json_event format. Options can also be set as environment variables.

NOTE: the redis transport uses a namespace of logstash:beaver by default. You will need to update your logstash indexer to match this, or you may configure beaver to do otherwise.

Configuration File Options

Beaver can optionally get data from a configfile using the -c flag. This file is in ini format. Global configuration will be under the beaver stanza. The following are global beaver configuration keys with their respective meanings:

• rabbitmq_host: Defaults localhost. Host for RabbitMQ.
• rabbitmq_port: Defaults 5672. Port for RabbitMQ.
• rabbitmq_vhost: Default /
• rabbitmq_username: Default guest
• rabbitmq_password: Default guest
• rabbitmq_queue: Default logstash-queue.
• rabbitmq_exchange: Default direct.
• rabbitmq_exchange_durable: Default 0.
• rabbitmq_key: Default logstash-key.
• rabbitmq_exchange: Default logstash-exchange.
• redis_url: Default redis://localhost:6379/0. Redis URL
• redis_namespace: Default logstash:beaver. Redis key namespace
• udp_host: Default 127.0.0.1. UDP Host
• udp_port: Default 9999. UDP Port
• zeromq_address: Default tcp://localhost:2120. Zeromq URL
• zeromq_bind: Default bind. Whether to bind to zeromq host or simply connect

The following are used for instances when a TransportException is thrown - Transport dependent

• respawn_delay: Default 3. Initial respawn delay for exponential backoff
• max_failure: Default 7. Max failures before exponential backoff terminates

The following configuration keys are for building an SSH Tunnel that can be used to proxy from the current host to a desired server. This proxy is torn down when Beaver halts in all cases.

• ssh_key_file: Default None. Full path to id_rsa key file
• ssh_tunnel: Default None. SSH Tunnel in the format user@host:port
• ssh_tunnel_port: Default None. Local port for SSH Tunnel
• ssh_remote_host: Default None. Remote host to connect to within SSH Tunnel
• ssh_remote_port: Default None. Remote port to connect to within SSH Tunnel

The following can also be passed via argparse. Argparse will override all options in the configfile, when specified.

• format: Default json. Options [ json, msgpack, string ]. Format to use when sending to transport
• files: Default files. Space-separated list of files to tail.
• path: Default /var/log. Path glob to tail.
• transport: Default stdout. Transport to use when log changes are detected
• fqdn: Default False. Whether to use the machine's FQDN in transport output
• hostname: Default None. Manually specified hostname

Examples

Example 1: Listen to all files in the default path of /var/log on standard out as json:

beaver

Example 2: Listen to all files in the default path of /var/log on standard out with msgpack:

beaver --format msgpack

Example 3: Listen to all files in the default path of /var/log on standard out as a string:

beaver --format string

Example 4: Sending logs from /var/log files to a redis list:

# /etc/beaver.conf
[beaver]
redis_url: redis://localhost:6379/0

# From the commandline
beaver  -c /etc/beaver.conf -t redis

Example 5: Use environment variables to send logs from /var/log files to a redis list:

# /etc/beaver.conf
[beaver]
redis_url: redis://localhost:6379/0

# From the commandline
beaver  -c /etc/beaver.conf -p '/var/log' -t redis

Example 6: Zeromq listening on port 5556 (all interfaces):

# /etc/beaver.conf
[beaver]
zeromq_address: tcp://*:5556

# logstash indexer config:
input {
  zeromq {
    type => 'shipper-input'
    mode => 'client'
    topology => 'pushpull'
    address => 'tcp://shipperhost:5556'
  }
}
output { stdout { debug => true } }

# From the commandline
beaver  -c /etc/beaver.conf -m bind -t zmq

Example 7: Zeromq connecting to remote port 5556 on indexer:

# /etc/beaver.conf
[beaver]
zeromq_address: tcp://indexer:5556

# logstash indexer config:
input {
  zeromq {
    type => 'shipper-input'
    mode => 'server'
    topology => 'pushpull'
    address => 'tcp://*:5556'
  }
}
output { stdout { debug => true } }

# on the commandline
beaver -c /etc/beaver.conf -m connect -t zmq

Example 8: Real-world usage of Redis as a transport:

# in /etc/hosts
192.168.0.10 redis-internal

# /etc/beaver.conf
[beaver]
redis_url: redis://redis-internal:6379/0
redis_namespace: app:unmappable

# logstash indexer config:
input {
  redis {
    host => 'redis-internal'
    data_type => 'list'
    key => 'app:unmappable'
    type => 'app:unmappable'
  }
}
output { stdout { debug => true } }

# From the commandline
beaver -c /etc/beaver.conf -f /var/log/unmappable.log -t redis

As you can see, beaver is pretty flexible as to how you can use/abuse it in production.

Example 9: RabbitMQ connecting to defaults on remote broker:

# /etc/beaver.conf
[beaver]
rabbitmq_host: 10.0.0.1

# logstash indexer config:
input { amqp {
    name => 'logstash-queue'
    type => 'direct'
    host => '10.0.0.1'
    exchange => 'logstash-exchange'
    key => 'logstash-key'
    exclusive => false
    durable => false
    auto_delete => false
  }
}
output { stdout { debug => true } }

# From the commandline
beaver -c /etc/beaver.conf -t rabbitmq

Example 10: Read config from config.ini and put to stdout:

# /etc/beaver.conf:
[/tmp/somefile]
type: mytype
tags: tag1,tag2
add_field: fieldname1,fieldvalue1[,fieldname2,fieldvalue2, ...]

[/var/log/*log]
type: syslog
tags: sys

[/var/log/{secure,messages}.log]
type: syslog
tags: sys

# From the commandline
beaver -c /etc/beaver.conf -t stdout

Example 11: UDP transport:

# /etc/beaver.conf
[beaver]
udp_host: 127.0.0.1
udp_port: 9999

# logstash indexer config:
input {
  udp {
    type => 'shipper-input'
    host => '127.0.0.1'
    port => '9999'
  }
}
output { stdout { debug => true } }

# From the commandline
beaver -c /etc/beaver.conf -t udp

Todo

• More documentation
• Use python threading + subprocess in order to support usage of yield across all operating systems
• ~~Fix usage on non-linux platforms - file.readline() does not work as expected on OS X. See above for potential solution~~
• More transports
• ~~Ability to specify files, tags, and other metadata within a configuration file~~

Caveats

When using copytruncate style log rotation, two race conditions can occur:

1. Any log data written prior to truncation which beaver has not yet read and processed is lost. Nothing we can do about that.

2. Should the file be truncated, rewritten, and end up being larger than the original file during the sleep interval, beaver won't detect this. After some experimentation, this behavior also exists in GNU tail, so I'm going to call this a "don't do that then" bug :)

   Additionally, the files beaver will most likely be called upon to watch which may be truncated are generally going to be large enough and slow-filling enough that this won't crop up in the wild.

Credits

Based on work from Giampaolo and Lusis:

Real time log files watcher supporting log rotation.

Original Author: Giampaolo Rodola' <g.rodola [AT] gmail [DOT] com>
http://code.activestate.com/recipes/577968-log-watcher-tail-f-log/

License: MIT

Other hacks (ZMQ, JSON, optparse, ...): lusis