fluent-plugin-postgres, a plugin for Fluentd

Installation:

Prereq: Install postgresql headers: apt-get install libpq-dev
Install the gem:
- gem install fluent-plugin-postgres or
- /usr/lib/fluent/ruby/bin/fluent-gem install fluent-plugin-postgres

Changes from mysql:

We currently don't support json format
You need to specify a SQL query
Placeholders are numbered (yeah, I know).

Other than that, just bear in mind that it's Postgres SQL.

Quick example

<match output.by.sql.*>
  type postgres
  host master.db.service.local
  # port 3306 # default
  database application_logs
  username myuser
  password mypass
  key_names status,bytes,vhost,path,rhost,agent,referer
  sql INSERT INTO accesslog (status,bytes,vhost,path,rhost,agent,referer) VALUES ($1,$2,$3,$4,$5,$6,$7)
  flush_intervals 5s
</match>

Component

PostgresOutput

Plugin to store Postgres tables over SQL, to each columns per values, or to single column as json.

Configuration

MysqlOutput

MysqlOutput needs MySQL server's host/port/database/username/password, and INSERT format as SQL, or as table name and columns.

<match output.by.sql.*>
  type mysql
  host master.db.service.local
  # port 3306 # default
  database application_logs
  username myuser
  password mypass
  key_names status,bytes,vhost,path,rhost,agent,referer
  sql INSERT INTO accesslog (status,bytes,vhost,path,rhost,agent,referer) VALUES (?,?,?,?,?,?,?)
  flush_intervals 5s
</match>

<match output.by.names.*>
  type mysql
  host master.db.service.local
  database application_logs
  username myuser
  password mypass
  key_names status,bytes,vhost,path,rhost,agent,referer
  table accesslog
  # 'columns' names order must be same with 'key_names'
  columns status,bytes,vhost,path,rhost,agent,referer
  flush_intervals 5s
</match>

Or, insert json into single column.

<match output.as.json.*>
  type mysql
  host master.db.service.local
  database application_logs
  username root
  table accesslog
  columns jsondata
  format json
  flush_intervals 5s
</match>

To include time/tag into output, use include_time_key and include_tag_key, like this:

<match output.with.tag.and.time.*>
  type mysql
  host my.mysql.local
  database anydatabase
  username yourusername
  password secret

  include_time_key yes
  ### default `time_format` is ISO-8601
  # time_format %Y%m%d-%H%M%S
  ### default `time_key` is 'time'
  # time_key timekey

  include_tag_key yes
  ### default `tag_key` is 'tag'
  # tag_key tagkey

  table anydata
  key_names time,tag,field1,field2,field3,field4
  sql INSERT INTO baz (coltime,coltag,col1,col2,col3,col4) VALUES (?,?,?,?,?,?)
</match>

Or, for json:

<match output.with.tag.and.time.as.json.*>
  type mysql
  host database.local
  database foo
  username root

  include_time_key yes
  utc   # with UTC timezome output (default: localtime)
  time_format %Y%m%d-%H%M%S
  time_key timeattr

  include_tag_key yes
  tag_key tagattr
  table accesslog
  columns jsondata
  format json
</match>
#=> inserted json data into column 'jsondata' with addtional attribute 'timeattr' and 'tagattr'

TODO

implement 'tag_mapped'
- dynamic tag based table selection

Copyright

License
- Apache License, Version 2.0

fluent-plugin-postgres

Development

Runtime