skip to navigation
skip to content

Beaver 30

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

Latest Version: 36.2.1

.. image::

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


* 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``)


Using PIP:

From Github::

pip install git+git://

From PyPI::

pip install beaver==30



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

optional arguments::

-h, --help show this help message and exit
-c CONFIG, --configfile CONFIG
main beaver ini config file path
-C CONFD_PATH ini config directory 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,raw,rawjson,string}, --format {json,msgpack,raw,rawjson,string}
format to use when sending to transport
manual hostname override for source_host
-m {bind,connect}, --mode {bind,connect}
bind or connect mode
-l OUTPUT, --logfile OUTPUT, -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 {mqtt,rabbitmq,redis,stdout,tcp,udp,zmq}, --transport {mqtt,rabbitmq,redis,sqs,stdout,tcp,udp,zmq}
log transport method
-v, --version output version and quit
--fqdn use the machine's FQDN for source_host


Beaver provides an lightweight method for shipping local log files to Logstash. It does this using redis, zeromq, tcp, udp, rabbit or stdout as the transport. This means you'll need a redis, zeromq, tcp, udp, amqp or stdin 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:

* mqtt_host: Default ``localhost``. Host for mosquitto
* mqtt_port: Default ``1883``. Port for mosquitto
* mqtt_clientid: Default ``mosquitto``. Mosquitto client id
* mqtt_keepalive: Default ``60``. mqtt keepalive ping
* mqtt_topic: Default ``/logstash``. Topic to publish to
* 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_type: 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
* sqs_aws_access_key: Can be left blank to use IAM Roles or AWS_ACCESS_KEY_ID environment variable (see:
* sqs_aws_secret_key: Can be left blank to use IAM Roles or AWS_SECRET_ACCESS_KEY environment variable (see:
* sqs_aws_region: Default ``us-east-1``. AWS Region
* sqs_aws_queue: SQS queue (must exist)
* tcp_host: Default ````. TCP Host
* tcp_port: Default ``9999``. TCP Port
* udp_host: Default ````. UDP Host
* udp_port: Default ``9999``. UDP Port
* zeromq_address: Default ``tcp://localhost:2120``. Zeromq URL
* zeromq_hwm: Default None. Zeromq HighWaterMark socket option
* 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 SinceDB support. Specifying these will enable saving the current line number in an sqlite database. This is useful for cases where you may be restarting the beaver process, such as during a logrotate.

* sincedb_path: Default ``None``. Full path to an ``sqlite3`` database. Will be created at this path if it does not exist. Beaver process must have read and write access

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. (Comma separated if specified in the config file)
* 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


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


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
redis_url: redis://localhost:6379/0

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

Example 5: Zeromq listening on port 5556 (all interfaces)::

# /etc/beaver/conf
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 6: Zeromq connecting to remote port 5556 on indexer::

# /etc/beaver/conf
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 7: Real-world usage of Redis as a transport::

# in /etc/hosts redis-internal

# /etc/beaver/conf
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

Example 8: RabbitMQ connecting to defaults on remote broker::

# /etc/beaver/conf

# logstash indexer config:
input { amqp {
name => 'logstash-queue'
type => 'direct'
host => ''
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 9: Read config from config.ini and put to stdout::

# /etc/beaver/conf:
; follow a single file, add a type, some tags and fields
type: mytype
tags: tag1,tag2
add_field: fieldname1,fieldvalue1[,fieldname2,fieldvalue2, ...]

; follow all logs in /var/log except those with `messages` or `secure` in the name.
; The exclude tag must be a valid python regular expression.
type: syslog
tags: sys
exclude: (messages|secure)

; follow /var/log/messages.log and /var/log/secure.log using file globbing
type: syslog
tags: sys

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

Example 10: TCP transport::

# /etc/beaver/conf
tcp_port: 9999

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

# From the commandline
beaver -c /etc/beaver/conf -t tcp

Example 11: UDP transport::

# /etc/beaver/conf
udp_port: 9999

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

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

Example 12: SQS Transport::

# /etc/beaver/conf
sqs_aws_region: us-east-1
sqs_aws_queue: logstash-input
sqs_aws_access_key: <access_key>
sqs_aws_secret_key: <secret_key>

# logstash indexer config:
input {
sqs {
queue => "logstash-input"
type => "shipper-input"
format => "json_event"
access_key => "<access_key>"
secret_key => "<secret_key>"
output { stdout { debug => true } }

# From the commandline
beaver -c /etc/beaver/conf -t sqs

Example 13: [Raw Json Support](

beaver --format rawjson

Example 14: Mqtt transport using Mosquitto::

# /etc/beaver/conf
mqtt_client_id: 'beaver_client'
mqtt_topic: '/logstash'
mqtt_host: ''
mqtt_port: '1318'
mqtt_keepalive: '60'

# logstash indexer config:
input {
mqtt {
host => ''
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 mqtt

Example 15: Sincedb support using and sqlite3 db

Note that this will require R/W permissions on the file at sincedb path, as Beaver will store the current line for a given filename/file id.::

# /etc/beaver/conf
sincedb_path: /etc/beaver/since.db

type: syslog
tags: sys,main
sincedb_write_interval: 3 ; time in seconds

# From the commandline
beaver -c /etc/beaver/conf

Example 16: Loading stanzas from /etc/beaver/conf.d/* support::

# /etc/beaver/conf
format: json

# /etc/beaver/conf.d/syslog
type: syslog
tags: sys,main

# /etc/beaver/conf.d/nginx
format: rawjson
type: nginx
tags: nginx,server

# From the commandline
beaver -c /etc/beaver/conf -C /etc/beaver/conf.d

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


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


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.

When you get an error similar to ``ImportError: No module named
_sqlite3`` your python seems to miss the sqlite3-module. This can be the
case on FreeBSD and probably other systems. If so, use the local package
manager or port system to build that module. On FreeBSD::

cd /usr/ports/databases/py-sqlite3
sudo make install clean


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="">

License: MIT

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


30 (2013-08-22)

- Use os._exit over sys.exit in signal handlers to quit cleanly.
[Kristian Glass]

As per
the use of `sys.exit` inside the signal handlers means that a
`SystemExit` exception is raised
( which can be
by try/except blocks that might have been executing at time of signal
handling, resulting in beaver failing to quit

- Allow path to be None. [Lars Hansson]

Allow path to be set empty (None) in the configuration filer. This way
all files and globs can be configured in files in confd_path.

- Fixed example in Readme.rst for sqs_aws_secret_key. [Jonathan Quail]

- Redis 2.4.11 is no longer available on Pypi. [Andrew Gross]

Fixes issue #167

- Add a TCP transport. [Kiall Mac Innes]

- Allow string escapes in delimiter. [Michael Mittelstadt]

As far as I can tell, there is no way for me to represent a newline as
a delimiter in a configuration file with ConfigParser. I want to do

tags: special
type: special
delimiter: \n\n

As the log has a blank line between its multiline entries.

My change allows that, by making delimiter not string
escaped until
after the config file is parsed. I'm naive about python, so there is a
strong possibility I've gone about it horribly wrong. This would also
easily allow splitting on nulls, tabs, unicode characters and other
things that ConfigParser may not find kosher.

By doing this sort of multiline parsing with beaver, it allows one to
run logstash without the multiline filter, which due to its lack of
safety, forces you to run logstash with only one worker thread.

- Isolate connection logic. Provide proper reconnect support. [Chris

- Fix zmq transport tests. [Scott Smith]

- Move zmq address config parsing into _main_parser. [Scott Smith]

- Allow specifying multiple zmq addresses to bind/connect. [Scott Smith]

- Corrected documentation for exclude tag. Closes #157. [Jose Diaz-

- Add missing sqlite3 module to documentation. [Andreas Lappe]

- Tests status. [Denis Orlikhin]

- Travis integration. [Denis Orlikhin]

- Tests fix (conf_d does work without existing file) [Denis Orlikhin]

- CONFIG_DIR to CONFD_PATH. [iyingchi]

- Added doc for -C option for config directory. [iyingchi]

- Implicit broken zmq error handling. [Denis Orlikhin]

29 (2013-05-24)

- Do not harcode path in TailManager. Closes #143. [Jose Diaz-Gonzalez]

- Use /etc/beaver/conf for path and provide conf.d example. Closes #149.
[Jose Diaz-Gonzalez]

- Added mqtt as option in argparse configuration for the transport flag.
[Jose Diaz-Gonzalez]

- Fixed broken MqttTransport naming. [Jose Diaz-Gonzalez]

- Refactored BeaverSubprocess to maintain the running command as an
attribute. [Jose Diaz-Gonzalez]

- Properly parse the beaver conf.d path for new sections. Closes #144.
Closes #145. Refs #107. [Jose Diaz-Gonzalez]

- Use a Buffered Tokenizer to read large/fast incoming log input. Refs
#135. Refs #105. [Jose Diaz-Gonzalez]

- Close queue after worker has been stopped. Refs #135. [Jose Diaz-

- Wrap manager.close() call in try/except to mimic the worker
dispatcher. [Jose Diaz-Gonzalez]

- Properly parse out the port from the `ssh_tunnel` option. Closes #142.
[Jose Diaz-Gonzalez]

- Subclass the BaseLog class in BeaverSubprocess. Refs #142. [Jose Diaz-

- Move base_log module higher up in hierarchy. Refs #142. [Jose Diaz-

- Disable daemonization on the windows platform. Closes #141. [Jose

- Move file unwatching in old-style worker out of for-loop. Refs #139.
[Jose Diaz-Gonzalez]

Each worker has a `self._file_map` attribute which is a mapping of
file ids to file data. When retrieving lines or checking on the status
of the file, we use `iteritems()` which gives us a generator as
opposed to a copy such as with `items()`. This generator allows us to
iterate over the files without having issues where the file handle may
open several times or other random Python issues.

Using a generator also means that the set that we are iterating over
should not change mid
iteration, which it does if a file is unwatched. To circumvent this,
we should use a separate list to keep track of files we need to
unwatch or rewatch, and do it out of band.

We should also take care to catch `RuntimeError` which may arise when
closing the Worker out of band
such as in the `cleanup` step of the worker dispatcher
but nowhere else.

This should fix issues where logrotate suddenly causes files to
disappear for a time and beaver tries to tail the file at the exact
time it is being recreated.

- Typo in SQS docs. [Jonathan Quail]

- Remove ujson requirement. [Jose Diaz-Gonzalez]

This allows users that do not have a compiler in their deployment area
to install beaver.

Closes #137

- Turn on logfile output when running in non-daemon contexts. Closes
#131. [Jose Diaz-Gonzalez]

- Expand logging output path. Closes #133. [Jose Diaz-Gonzalez]

- Ensure logging to a file does not destroy regular logging. Closes
#132. [Jose Diaz-Gonzalez]

- Properly handle unreadable files by logging a warning instead of
crashing. Closes #130. [Jose Diaz-Gonzalez]

- Rename null_formatter to raw_formatter in BaseTransport class. [Jose

- Ensure that the RedisTransport calls the super invalidate method. Refs
#93. [Jose Diaz-Gonzalez]

- Fix issue where input type was not being detected properly. [Jose

- Use logfile flag for sending all output to a file in daemon contexts.
[Jose Diaz-Gonzalez]

- Expand path for pidfile creation. [Jose Diaz-Gonzalez]

- Properly handle redis reconnects when the datastore becomes
unreacheable. Refs #93. [Jose Diaz-Gonzalez]

- 'type' instead of 'exchange_type' in recent pika vers. [Pravir

- Adding options to make queues durable and HA. [Pravir Chandra]

- Respect stat_interval file configuration in stable worker. [Jose Diaz-

- Unified configuration file using conf_d module. [Jose Diaz-Gonzalez]

This change adds support for a conf.d directory
configured only via the `

path` flag
which allows beaver to read configuration from multiple files.

Please note that the primary `beaver` stanza MUST be located in the
file specified by the `

configfile` argument. Any other such `beaver` stanzas will be ignored.

This change also unifies the `BeaverConfig` and `FileConfig` classes,
and simplifies the api for retrieving global vs file
specific data.

Please note that this commit BREAKS custom transport classes, as the
interface for creating a transport class has changed. If you are
referencing a `file_config.get(field, filename)` anywhere, please omit
this and refer to `beaver_config.get_field(field, filename)`.

Closes #107

- Hack to prevent stupid TypeError: 'NoneType' when running tests via [Jose Diaz-Gonzalez]

- Properly handle rotated files on Darwin architectures. [Jose Diaz-

- Log to debug instead of warning for file reloading on Darwin
architectures. [Jose Diaz-Gonzalez]

- Speed up experimental worker. [Jose Diaz-Gonzalez]

Removed inline sleep call, which slowed down passes n*0.1 seconds,
where n is the number of files being tailed

Inline methods that update data structures which should speed up
larger installations

Make an attribute lookup instead of a method call

- Use latest version of message pack interface (0.3.0). Closes #128.
[Jose Diaz-Gonzalez]

- Alternative for reading python requirements. [Justin Lambert]

- Fix options sent from original worker to queue. Refs #119. [Jose Diaz-

- Allow users to ignore the results of a copytruncate from logrotate.
Refs #105. [Jose Diaz-Gonzalez]

- Fix rpm package building. Closes #123. [Jose Diaz-Gonzalez]

- Added experimental tail-version of beaver. [Jose Diaz-Gonzalez]

- Beginning work to move from an omniscient worker to individual tail
objects. [Jose Diaz-Gonzalez]

- Fix kwargs call. [Jose Diaz-Gonzalez]

- Add formatting to mqtt transport. Closes #115. [Jose Diaz-Gonzalez]

- Retrieve more data from callback to minimize dictionary lookups. [Jose

- Prefer single quotes to double quotes where possible. [Jose Diaz-

- Ensure stat_interval and tail_lines are both integer values. [Jose

- Alphabetize config variables for file_config. [Jose Diaz-Gonzalez]

- Ensure that debug flag is a boolean. [Jose Diaz-Gonzalez]

- Follow logstash covention for 'format' instead of 'message_format'
[Jose Diaz-Gonzalez]

- Use passed in 'ignore_empty' field instead of a file_config lookup in
queue module. [Jose Diaz-Gonzalez]

- Prefer discover_interval over update_file_mapping_time. [Jose Diaz-

- Fix TransportException import. Closes #122. [Jose Diaz-Gonzalez]

- Use an alternative method of reading in requirements. Refs #120. [Jose

- Auto-reconnect mechanism for the SSH tunnel. [Michael Franz Aigner]

- Fix import of REOPEN_FILES constant in [Jose Diaz-

- Fix a PEP8 violation. [Jose Diaz-Gonzalez]

- Ensure all files are utf-8 encoded. [Jose Diaz-Gonzalez]

- Namespace transport classes in the transport module. [Jose Diaz-

- Allow specifying debug mode via argument. [Jose Diaz-Gonzalez]

- Added thread-safety to datetime calls. [Jose Diaz-Gonzalez]

- Added support for message_format. Closes #91. [Jose Diaz-Gonzalez]

- Add msgpack_pure as fallback for C-Based msgpack package. [Jose Diaz-

- Fix issues in sincedb implementation. Refs #116. [Jose Diaz-Gonzalez]

- Fix casting issue when checking start_position. [Jose Diaz-Gonzalez]

- Properly handle Queue.Full exceptions. [Jose Diaz-Gonzalez]

- More logging. [Jose Diaz-Gonzalez]

- Expand the sincedb path on configuration parse. [Jose Diaz-Gonzalez]

- Ignore since.db files. [Jose Diaz-Gonzalez]

- Simplified sincedb support to handle an edge case. Refs #116. [Jose

- Remove errant print. [Jose Diaz-Gonzalez]

- Added support for file exclusion in config stanzas. Closes #106. [Jose

- Added python regex exclusion support to eglob. Refs #106. [Jose Diaz-

- PEP8. [Jose Diaz-Gonzalez]

- Added a tests directory with some sample tests from users. [Jose Diaz-

- Convert the 'sincedb_write_interval' option to an integer. Refs #116.
[Jose Diaz-Gonzalez]

- Moved logger call to a more intelligent spot. [Jose Diaz-Gonzalez]

- Ensure that we use the proper encoding when opening a file. Closes
#104. [Jose Diaz-Gonzalez]

- Centralize file-reading using classmethod open() [Jose Diaz-Gonzalez]

- Fixed issue where tailed lines were not being properly sent to the
callback. [Jose Diaz-Gonzalez]

- Remove unnecessary argument from Worke.__init__() [Jose Diaz-Gonzalez]

- Force-parse non-unicode files using unicode_dammit. [Jose Diaz-

- Set utf-8 as default encoding on all python files. [Jose Diaz-

- Fixed pyflakes issues. [rtoma]

- Syntax fix of list. [rtoma]

- Raise an AssertionError when run in daemon without a pid path
specified. Closes #112. [Jose Diaz-Gonzalez]

- Add support for ignoring empty lines. [Jose Diaz-Gonzalez]

- Properly cast boolean values from strings. [Jose Diaz-Gonzalez]

- Ensure all sections have the proper values on start. [Jose Diaz-

- Ensure internal file_config state is updated. [Jose Diaz-Gonzalez]

- Pass in timestamp from worker class for more accurate timestamps at
the cost of speed of sending. [Jose Diaz-Gonzalez]

- Centralize timestamp retrieval to base transport class. [Jose Diaz-

- Added support for gzipped files. refs #39. [Jose Diaz-Gonzalez]

- Added support for sqlite3-based sincedb. Refs #6 and #39. [Jose Diaz-

- Refactored worker so as to allow further data to be added to the
file_map. [Jose Diaz-Gonzalez]

- Refactor seek_to_end to properly support file tailing. [Jose Diaz-

- Added support for pubsub zmq. [Jose Diaz-Gonzalez]

- Added support for mosquitto transport. [Jose Diaz-Gonzalez]

- Added support for specifying file encoding, using vs
[Jose Diaz-Gonzalez]

- Fix issue where a field may not exist in the data. [Jose Diaz-

- Added support for rawjson format. [Jose Diaz-Gonzalez]

- Fixed zeromq tests. [Jose Diaz-Gonzalez]

- Added SQS transport. [Jonathan Quail]

- Fixing outdated transport docs. [Morgan Delagrange]

28 (2013-03-05)

- BeaverSubprocess is now a new-style class. Fixes ssh_tunneling. [Jose

27 (2013-03-05)

- Fix issue where super method was not called in BeaverSshTunnel. [Jose

26 (2013-03-05)

- Add optional reconnect support for transports. Refs #93. [Jose Diaz-

- Add a method for checking the validity of a Transport. Refs #93. [Jose

- Added a configurable subprocess poll sleep. [Jose Diaz-Gonzalez]

- Add a deafult sleep timeout to BeaverSubprocess polling. [Jose Diaz-

- Use a larger sleep time to get around redis over ssh connection
issues. [Jose Diaz-Gonzalez]

25 (2013-03-05)

- Use True instead of 1 for while check. [Jose Diaz-Gonzalez]

- Fix orphan child processes. Closes #103. [Jose Diaz-Gonzalez]

24 (2013-02-26)

- Ensure new files are added to a transports configuration. Closes #96.
Closes #101. [Jose Diaz-Gonzalez]

- Allow float numbers for update_file_mapping_time. [Jose Diaz-Gonzalez]

- Fix invalid casting of boolean values. [Jose Diaz-Gonzalez]

- Perform all conversions in Closes #99. [Jose Diaz-Gonzalez]

23 (2013-02-20)

- Worker: pretty format debug message "Iteration took %.6f" [Sergey

- Zeromq_hwm int() conversion moved to config. [Denis Orlikhin]

- Zeromq_hwm config entry. [Denis Orlikhin]

- Zeromq_hwm support. [Denis Orlikhin]

- Rabbitmq_exchange_type option fixed in the README. [Xabier de Zuazo]

- Add test requirements to setup. [Paul Garner]

- Allow beaver to accept custom transport classes. [Paul Garner]

- Make beaver slightly more amenable to test mocking and sort of fix the
broken zmq test. [Paul Garner]

22 (2013-01-15)

- Handle sigterm properly. Refs #87. [Jose Diaz-Gonzalez]

- Add --loglevel as alias for --output. Closes #92. [Jose Diaz-Gonzalez]

- Added logging on connection exception. [Thomas Morse]

- Add '--format raw' to pass through input unchanged. [Stephen Sugden]

- Fix string & null formatters in beaver.transport. [Stephen Sugden]

the inline definitions were expecting a self parameter, which is *not*
passed when you assign a function to an attribute on an object

- Adding exception when redis connection can't be confirmed. [William

- Call file.readlines() with sizehint in a loop to avoid reading in
massive files all at once. [Jose Diaz-Gonzalez]

21 (2013-01-04)

- Move runner into a dispatcher class to solve installation issues.
[Jose Diaz-Gonzalez]

- Added note for Python 2.6+ support. [Jose Diaz-Gonzalez]

20 (2013-01-03)

- Copy the readme over to avoid pypi packaging warnings. [Jose Diaz-

- Implement fully recursive file globing. [Brian L. Troutwine]

Python's base glob.iglob does not operate as if globstar were in
effect. To
explain, let's say I have an erlang application with lager logs to


and webmachine logs to


Prior to this commit, when configured with the path
`/var/log/**/*.log` all
webmachine logs would be ignored by beaver. This is no longer the
case, to an
arbitrary depth.

by: Brian L. Troutwine <>

19 (2013-01-01)

- Fix issue with supporting command line args. [Jose Diaz-Gonzalez]

18 (2012-12-31)

- Add timing debug information to the worker loop. [Jose Diaz-Gonzalez]

- Use redis pipelining when sending events. [Jose Diaz-Gonzalez]

- Formatting. [Jose Diaz-Gonzalez]

- Do not output debug statement for file_config.get call. [Jose Diaz-

- Pass in logger object to create_ssh_tunnel() [Jose Diaz-Gonzalez]

17 (2012-12-28)

- Added missing python-daemon requirement. [Jose Diaz-Gonzalez]

16 (2012-12-27)

- Specify a max queue size of 100 to limit overrunning memory. [Jose

- Use multiprocessing for handling larger queue sizes. [Jose Diaz-

Previously there were issues where files that were updated frequently
such as varnish or server logs
would overwhelm the naive implementation of file.readlines() within
Beaver. This would cause Beaver to slowly read larger and larger
portions of a file before processing any of the lines, eventually
causing Beaver to take forever to process log lines.

This patch adds the ability to use an internal work queue for log
lines. Whenever file.readlines() is called, the lines are placed in
the queue, which is shared with a child process. The child process
creates its own transport, allowing us to potentially create a Process
Pool in the future to handle a larger queue size.

Note that the limitation of file.readlines() reading in too many lines
is still in existence, and may continue to cause issues for certain
log files.

- Add default redis_password to BeaverConfig class. [Jose Diaz-Gonzalez]

- Fix missing underscore causing transport to break. [Norman Joyner]

- Implement redis auth support. [Norman Joyner]

- Add beaver init script for daemonization mode. [Jose Diaz-Gonzalez]

- Use python logger when using StdoutTransport. [Jose Diaz-Gonzalez]

- Add short arg flags for hostname and format. [Jose Diaz-Gonzalez]

- Add the ability to daemonize. Closes #79. [Jose Diaz-Gonzalez]

- Pass around a logger instance to all transports. [Jose Diaz-Gonzalez]

- Revert "Added a lightweight Event class" [Jose Diaz-Gonzalez]

After deliberation, beaver is meant to be "light
weight". Lets leave
the heavy
hitting to the big

This reverts commit 1619d33ef4803c3fe910cf4ff197d0dd0039d2eb.

- Added a lightweight Event class. [Jose Diaz-Gonzalez]

This class's sole responsibility will be the processing of a given
line as an event.
It's future goal will be to act as a lightweight implementation of the
filter system within Logstash

- Remove argparse requirement for python 2.7 and above. [Jose Diaz-

15 (2012-12-25)

- Pull argument parsing out of beaver [Jose Diaz-Gonzalez]

- Move app-running into [Jose Diaz-Gonzalez]

- Standardize on _parse() as method for parsing config. [Jose Diaz-

- Automatically parse the path config option. [Jose Diaz-Gonzalez]

- Remove extensions argument on Worker class. [Jose Diaz-Gonzalez]

This argument was only used when no globs were specified in a config
Since it is not configurable, there is no sense leaving around the
extra logic.

- Remove extra callback invocation on readlines. [Jose Diaz-Gonzalez]

- Remove extra file_config module. [Jose Diaz-Gonzalez]

- General code reorganization. [Jose Diaz-Gonzalez]

Move both BeaverConfig and FileConfig into a single class

Consolidated run_worker code with code in beaver binary file. This
will create a clearer path for Exception handling, as it is now the
responsibility of the calling class, allowing us to remove duplicative
exception handling code.

Added docstrings to many fuctions and methods

Moved extra configuration and setup code to beaver.utils module. In
many cases, code was added hastily before.

Made many logger calls debug as opposed to info. The info level should
be generally reserved for instances where files are watched,
unwatched, or some change in the file state has occurred.

- Remove duplicative and old beaver instructions from binary. [Jose

- Remove unnecessary passing of ssh_tunnel subprocess. [Jose Diaz-

- Added docstrings to ssh_tunnel module. [Jose Diaz-Gonzalez]

- Follow convention of underscore for object properties. [Jose Diaz-

- Follow convention of underscore for object properties. [Jose Diaz-

- Added a NullFormatter. [Jose Diaz-Gonzalez]

Useful for cases where we do not want any extra overhead on message

- Refactored message formatting in base Transport class. [Jose Diaz-

We now use a `_formatter` property on the Transport class which
will properly process the message for output as the user expects.

In the case of string output, we define a custom formatter using an
anonymous function and specify that as the formatter.

- Moved create_transport to transport module. [Jose Diaz-Gonzalez]

- Moved create_ssh_tunnel to ssh_tunnel module. [Jose Diaz-Gonzalez]

- Fixed order of beaver_config and file_config in args. [Jose Diaz-

- Reduce overhead of parsing configuration for globs and files. [Jose

- Removed ordereddict dependency. [Jose Diaz-Gonzalez]

- Do not output info level when outputing version. [Jose Diaz-Gonzalez]

- Allow usage of ujson >= 1.19. Closes #76. [Jose Diaz-Gonzalez]

14 (2012-12-18)

- Removed erroneous redundant code. [Jose Diaz-Gonzalez]

- Workaround for differing iteration implementation in Python 2.6. [Jose

- Properly detect non-linux platforms. [Jose Diaz-Gonzalez]

- Improve Python 2.6 support. [Jose Diaz-Gonzalez]

- Fix broken python readme. [Jose Diaz-Gonzalez]

13 (2012-12-17)

- Fixed certain environment variables. [Jose Diaz-Gonzalez]

- SSH Tunnel Support. [Jose Diaz-Gonzalez]

This code should allow us to create an ssh tunnel between two distinct
servers for the purposes of sending and receiving data.

This is useful in certain cases where you would otherwise need to
whitelist in your Firewall or iptables setup, such as when running in
two different regions on AWS.

- Allow for initial connection lag. Helpful when waiting for an SSH
proxy to connect. [Jose Diaz-Gonzalez]

- Fix issue where certain config defaults were of an improper value.
[Jose Diaz-Gonzalez]

- Allow specifying host via flag. Closes #70. [Jose Diaz-Gonzalez]

12 (2012-12-17)

- Reload tailed files on non-linux platforms. [Jose Diaz-Gonzalez]

Python has an issue on OS X were the underlying C implementation of
`` caches the EOF, therefore causing `readlines()` to only
work once. This happens to also fail miserably when you are seeking to
the end before calling readlines.

This fix solves the issue by constantly re
reading the files changed.

Note that this also causes debug mode to be very noisy on OS X. We all
have to make sacrifices...

- Deprecate all environment variables. [Jose Diaz-Gonzalez]

This shifts configuration management into the BeaverConfig class.
Note that we currently throw a warning if you are using environment

Refs #72
Closes #60

- Warn when using deprecated ENV variables for configuration. Refs #72.
[Jose Diaz-Gonzalez]

- Minor changes for PEP8 conformance. [Jose Diaz-Gonzalez]

11 (2012-12-16)

- Add optional support for socket.getfqdn. [Jeremy Kitchen]

For my setup I need to have the fqdn used at all times since my
hostnames are the same but the environment (among other things) is
found in the rest of the FQDN.

Since just changing socket.gethostname to socket.getfqdn has lots of
potential for breakage, and socket.gethostname doesn't always return
FQDN, it's now an option to explicitly always use the fqdn.

Fixes #68

- Check for log file truncation fixes #55. [Jeremy Kitchen]

This adds a simple check for log file truncation and resets the watch
when detected.

There do exist 2 race conditions here:
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
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.

- Add a version number to beaver. [Jose Diaz-Gonzalez]

10 (2012-12-15)

- Fixed package name. [Jose Diaz-Gonzalez]

- Regenerate CHANGES.rst on release. [Jose Diaz-Gonzalez]

- Adding support for /path/{foo,bar}.log. [Josh Braegger]

- Ignore file errors in unwatch method -- the file might not exists.
[Josh Braegger]

- Unwatch file when encountering a stale NFS handle. When an NFS file
handle becomes stale (ie, file was removed), it was crashing beaver.
Need to just unwatch file. [Josh Braegger]

- Consistency. [Chris Faulkner]

- Pull install requirements from requirements/base.txt so they don't get
out of sync. [Chris Faulkner]

- Include changelog in setup. [Chris Faulkner]

- Convert changelog to RST. [Chris Faulkner]

- Actually show the license. [Chris Faulkner]

- Consistent casing. [Chris Faulkner]

- Consistency. [Chris Faulkner]

- Stating the obvious. [Chris Faulkner]

- Grist for the mill. [Chris Faulkner]

- Drop redundant README.txt. [Chris Faulkner]

- Don't use empty string for tag when no tags configured in config file.
[Stylianos Modes]

- Making 'mode' option work for zmqtransport. Adding setuptools and
tests (use ./ nosetests). Adding .gitignore. [Josh Braegger]

9 (2012-11-28)

- More release changes. [Jose Diaz-Gonzalez]

- Fixed deprecated warning when declaring exchange type. [Rafael

7 (2012-11-28)

- Added a helper script for creating releases. [Jose Diaz-Gonzalez]

- Partial fix for crashes caused by globbed files. [Jose Diaz-Gonzalez]

- Removed deprecated usage of e.message. [Rafael Fonseca]

- Fixed exception trapping code. [Rafael Fonseca]

- Added some resiliency code to rabbitmq transport. [Rafael Fonseca]

6 (2012-11-26)

- Fix issue where polling for files was done incorrectly. [Jose Diaz-

- Added ubuntu init.d example config. [Jose Diaz-Gonzalez]

5 (2012-11-26)

- Try to poll for files on startup instead of throwing exceptions.
Closes #45. [Jose Diaz-Gonzalez]

- Added python 2.6 to classifiers. [Jose Diaz-Gonzalez]

4 (2012-11-26)

- Remove unused local vars. [Jose Diaz-Gonzalez]

- Allow rabbitmq exchange type and durability to be configured. [Jose

- Remove unused import. [Jose Diaz-Gonzalez]

- Formatted code to fix PEP8 violations. [Jose Diaz-Gonzalez]

- Use alternate dict syntax for Python 2.6 support. Closes #43. [Jose

- Fixed release date for version 3. [Jose Diaz-Gonzalez]

3 (2012-11-25)

- Added requirements files to manifest. [Jose Diaz-Gonzalez]

- Include all contrib files in release. [Jose Diaz-Gonzalez]

- Revert "removed redundant README.txt" to follow pypi standards. [Jose

This reverts commit e667f63706e0af8bc82c0eac6eac43318144e107.

- Added bash startup script. Closes #35. [Jose Diaz-Gonzalez]

- Added an example supervisor config for redis. closes #34. [Jose Diaz-

- Removed redundant README.txt. [Jose Diaz-Gonzalez]

- Added classifiers to package. [Jose Diaz-Gonzalez]

- Re-order workers. [Jose Diaz-Gonzalez]

- Re-require pika. [Jose Diaz-Gonzalez]

- Make zeromq installation optional. [Morgan Delagrange]

- Formatting. [Jose Diaz-Gonzalez]

- Added changes to changelog for version 3. [Jose Diaz-Gonzalez]

- Timestamp in ISO 8601 format with the "Z" sufix to express UTC.
[Xabier de Zuazo]

- Adding udp support. [Morgan Delagrange]

- Lpush changed to rpush on redis transport. This is required to always
read the events in the correct order on the logstash side. See: https:
af4620/lib/logstash/outputs/redis.rb#L4. [Xabier de Zuazo]

2 (2012-10-25)

- Example upstart script. [Michael D'Auria]

- Fixed a few more import statements. [Jose Diaz-Gonzalez]

- Fixed binary call. [Jose Diaz-Gonzalez]

- Refactored logging. [Jose Diaz-Gonzalez]

- Improve logging. [Michael D'Auria]

- Removed unnecessary print statements. [Jose Diaz-Gonzalez]

- Add default stream handler when transport is stdout. Closes #26. [bear
(Mike Taylor)]

- Handle the case where the config file is not present. [Michael

- Better exception handling for unhandled exceptions. [Michael D'Auria]

- Fix wrong addfield values. [Alexander Fortin]

- Add add_field to config example. [Alexander Fortin]

- Add support for add_field into config file. [Alexander Fortin]

- Minor readme updates. [Jose Diaz-Gonzalez]

- Add support for type reading from INI config file. [Alexander Fortin]

Add support for symlinks in config file

Add support for file globbing in config file

Add support for tags

a little bit of refactoring, move type and tags check down into
transport class

create config object (reading /dev/null) even if no config file
has been given via cli

Add documentation for INI file to readme

Remove unused json library


- When sending data over the wire, use UTC timestamps. [Darren Worrall]

- Support globs in file paths. [Darren Worrall]

- Added msgpack support. [Jose Diaz-Gonzalez]

- Use the python logging framework. [Jose Diaz-Gonzalez]

- Fixed Transport.format() method. [Jose Diaz-Gonzalez]

- Properly parse BEAVER_FILES env var. [Jose Diaz-Gonzalez]

- Refactor transports. [Jose Diaz-Gonzalez]

Fix the json import to use the fastest json module available

Move formatting into Transport class

- Attempt to fix defaults from env variables. [Jose Diaz-Gonzalez]

- Fix README and beaver CLI help to reference correct RABBITMQ_HOST
environment variable. [jdutton]

- Add RabbitMQ support. [Alexander Fortin]

- Added real-world example of beaver usage for tailing a file. [Jose

- Removed unused argument. [Jose Diaz-Gonzalez]

- Ensure that python-compatible readme is included in package. [Jose

- Fix variable naming and timeout for redis transport. [Jose Diaz-

- Installation instructions. [Jose Diaz-Gonzalez]

- Use restructured text for readme instead of markdown. [Jose Diaz-

- Removed unnecessary .gitignore. [Jose Diaz-Gonzalez]

1 (2012-08-06)

- Moved app into python package format. [Jose Diaz-Gonzalez]

- Moved binary to bin/beaver, as per python packaging. [Jose

- Moved around transports to be independent of each other. [Jose Diaz-

- Reorder transports. [Jose Diaz-Gonzalez]

- Rewrote run_worker to throw exception if all transport options have
been exhausted. [Jose Diaz-Gonzalez]

- Rename Amqp -> Zmq to avoid confusion with RabbitMQ. [Alexander

- Added choices to the --transport argument. [Jose Diaz-Gonzalez]

- Fixed derpy formatting. [Jose Diaz-Gonzalez]

- Added usage to the readme. [Jose Diaz-Gonzalez]

- Support usage of environment variables instead of arguments. [Jose

- Fixed files argument parsing. [Jose Diaz-Gonzalez]

- One does not simply license all the things. [Jose Diaz-Gonzalez]

- Add todo to readme. [Jose Diaz-Gonzalez]

- Added version to pyzmq. [Jose Diaz-Gonzalez]

- Added license. [Jose Diaz-Gonzalez]

- Reordered imports. [Jose Diaz-Gonzalez]

- Moved all transports to beaver/ [Jose Diaz-Gonzalez]

- Calculate current timestamp at most once per callback fired. [Jose

- Modified transports to include proper information for ingestion in
logstash. [Jose Diaz-Gonzalez]

- Fixed package imports. [Jose Diaz-Gonzalez]

- Removed another compiled python file. [Jose Diaz-Gonzalez]

- Use ujson instead of simplejson. [Jose Diaz-Gonzalez]

- Ignore compiled python files. [Jose Diaz-Gonzalez]

- Fixed imports. [Jose Diaz-Gonzalez]

- Fixed up readme instructions. [Jose Diaz-Gonzalez]

- Refactor transports so that connections are no longer global. [Jose

- Readme and License. [Jose Diaz-Gonzalez]  
File Type Py Version Uploaded on Size
Beaver-30.tar.gz (md5) Source 2013-08-22 72KB