flosculus

Tail your log, extract the data, and send it to Fluentd


License
MIT
Install
pip install flosculus==0.3.0

Documentation

Flosculus

Flosculus is a work-in-progress script to tail rotated log file, parse each line, and send it to Fluentd. It is intended to replace a subset of Fluentd in_tail features.

Crash Course

First things first, install Flosculus into your Python path.

$ git clone git://github.com/iromli/flosculus.git
$ cd flosculus
$ python setup.py install

Afterwards you need to create a configuration file to run the flosculusd event loop. The convenient way to do this boring thing is by invoking the flosculusd --init > flosculus.ini command. Here's an example of flosculus.ini (you can call it whatever you like):

[flosculus]
; the IP address (or host name) of the remote server
remote_host = 127.0.0.1

; the TCP port of the remote server
remote_port = 24224


; Each section with `path:/path/to/log` is a valid config
[log:/var/log/nginx/access.log]

; the label
tag = example.api.access

; format to use, either use 'nginx' or custom regex
format = nginx

By default, the format option is matched against nginx default access log format. You may change the format as long as using a valid Python regex.

[log:/var/log/nginx/access.log]

format = (?P<remote>[^ ]*) (?P<host>[^ ]*) (?P<user>[^ ]*) \[(?P<time>[^\]]*)\] "(?P<method>\S+)(?: +(?P<path>[^\"]*) +\S*)?" (?P<code>[^ ]*) (?P<size>[^ ]*)(?: "(?P<referer>[^\"]*)" "(?P<agent>[^\"]*)")(?: (?P<request_time>[^ ]*) (?P<upstream_time>[^ ]*) (?P<pipe>[\.|p]))?

And did I already mentioned that you can have multiple log files? Simply copy the whole log section.

; Each section with `path:/path/to/log` is a valid config
[log:/var/log/nginx/access.log]

; the label
tag = example.api.access

; format to use, either use 'nginx' or custom regex
format = nginx

; Each section with `path:/path/to/log` is a valid config
[log:/var/log/nginx/timed-combined.access.log]

; the label
tag = test.api.access

; format to use, either use 'nginx' or custom regex
format = (?P<remote>[^ ]*) (?P<host>[^ ]*) (?P<user>[^ ]*) \[(?P<time>[^\]]*)\] "(?P<method>\S+)(?: +(?P<path>[^\"]*) +\S*)?" (?P<code>[^ ]*) (?P<size>[^ ]*)(?: "(?P<referer>[^\"]*)" "(?P<agent>[^\"]*)")(?: (?P<request_time>[^ ]*) (?P<upstream_time>[^ ]*) (?P<pipe>[\.|p]))?

; the IP address (or host name) of the remote server
; use another Fluentd remote host
remote_host = 10.0.0.1

; the TCP port of the remote server
; use another Fluentd remote port
remote_port = 24225

Assuming the configuration is written properly, run the event loop to see its magic (well, not really):

$ flosculusd -c flosculus.ini

Credits

  • Beaver
  • Fluentd
  • Giampaolo Rodola for the original code (MIT-licensed) that evolved into flosculus.watcher module