MongoDB plugin for Fluentd

fluent-plugin-mongo provides input and output plugins for Fluentd (GitHub)

Installation

Gems

The gem is hosted at Rubygems.org. You can install the gem as follows:

$ fluent-gem install fluent-plugin-mongo

Plugins

Output plugin

mongo

Store Fluentd event to MongoDB database.

Configuration

Use mongo type in match.

<match mongo.**>
  type mongo
  database fluent
  collection test

  # Following attibutes are optional
  host fluenter
  port 10000

  # Set 'capped' if you want to use capped collection
  capped
  capped_size 100m

  # Set 'user' and 'password' for authentication
  user handa
  password shinobu

  # Other buffer configurations here
</match>

mongo(tag mapped mode)

Tag mapped to MongoDB collection automatically.

Configuration

Use tag_mapped parameter in match of mongo type.

If tag name is “foo.bar”, auto create collection “foo.bar” and insert data.

<match forward.*>
  type mongo
  database fluent

  # Set 'tag_mapped' if you want to use tag mapped mode.
  tag_mapped

  # If tag is "forward.foo.bar", then prefix "forward." is removed.
  # Collection name to insert is "foo.bar".
  remove_tag_prefix forward.

  # This configuration is used if tag not found. Default is 'untagged'.
  collection misc

  # Other configurations here
</match>

mongo_replset

Replica Set version of mongo.

Configuration

Use mongo_replset type in match.

<match mongo.**>
  type mongo_replset
  database fluent
  collection logs

  # each node separated by ','
  nodes localhost:27017,localhost:27018,localhost:27019

  # num_retries is threshold at failover, default is 60.
  # If retry count reached this threshold, mongo plugin raises an exception.
  num_retries 30

  # following optional parameters passed to ReplSetConnection of mongo-ruby-driver.
  # See mongo-ruby-driver docs for more detail.
  #name replset_name
  #read secondary
  #refresh_mode sync
  #refresh_interval 60
</match>

mongo_backup

Store Fluentd event to local capped collection for backup.

Configuration

Use mongo_backup type in match. mongo_backup alwalys use capped collection.

<match ...>
  type mongo_backup
  capped_size 100m

  <store>
    type tcp
    host 192.168.0.13
    ...
  </store>
</match>

Input plugin

mongo_tail

Tail capped collection to input data.

Configuration

Use mongo_tail type in source.

<source>
  type mongo_tail
  database fluent
  collection capped_log

  tag app.mongo_log

  # waiting time when there is no next document. default is 1s.
  wait_time 5

  # Convert 'time'(BSON's time) to fluent time(Unix time).
  time_key time

  # You can store last ObjectId to tail over server's shutdown
  id_store_file /Users/repeatedly/devel/fluent-plugin-mongo/last_id
</source>

NOTE

Broken data as a BSON

Fluentd event sometimes has an invalid record as a BSON. In such case, Mongo plugin marshals an invalid record using Marshal.dump and re-inserts its to same collection as a binary.

If passed following invalid record:

{"key1": "invalid value", "key2": "valid value", "time": ISODate("2012-01-15T21:09:53Z") }

then Mongo plugin converts this record to following format:

{"__broken_data": BinData(0, Marshal.dump result of {"key1": "invalid value", "key2": "valid value"}), "time": ISODate("2012-01-15T21:09:53Z") }

Mongo-Ruby-Driver cannot detect an invalid attribute, so Mongo plugin marshals all attributes excluding Fluentd keys(“tag_key” and “time_key”).

You can deserialize broken data using Mongo and Marshal.load. Sample code is below:

# _collection_ is an instance of Mongo::Collection
collection.find({'__broken_data' => {'$exists' => true}}).each do |doc|
  p Marshal.load(doc['__broken_data'].to_s) #=> {"key1": "invalid value", "key2": "valid value"}
end

ignore_invalid_record

If you want to ignore an invalid record, set true to ignore_invalid_record parameter in match.

<match forward.*>
  ...

  # ignore invalid documents at write operation
  ignore_invalid_record true

  ...
</match>

exclude_broken_fields

If you want to exclude some fields from broken data marshaling, use exclude_broken_fields to specfiy the keys.

<match forward.*>
  ...

  # key2 is excluded from __broken_data.
  # e.g. {"__broken_data": BinData(0, Marshal.dump result of {"key1": "invalid value"}), "key2": "valid value", "time": ISODate("2012-01-15T21:09:53Z")
  exclude_broken_fields key2

  ...
</match>

Specified value is a comma separated keys(e.g. key1,key2,key3). This parameter is useful for excluding shard keys in shard environment.

replace_dot_in_key_with and replace_dollar_in_key_with

BSON records which include ‘.’ or start with ‘$’ are invalid and they will be stored as broken data to MongoDB. If you want to sanitize keys, you can use replace_dot_in_key_with and replace_dollar_in_key_with.

<match forward.*>
  ...
  # replace '.' in keys with '__dot__'
  replace_dot_in_key_with __dot__

  # replace '$' in keys with '__dollar__'
  # Note: This replaces '$' only on first character
  replace_dollar_in_key_with __dollar__
  ...
</match>

Buffer size limitation

Mongo plugin has the limitation of buffer size. Because MongoDB and mongo-ruby-driver checks the total object size at each insertion. If total object size gets over the size limitation, then MongoDB returns error or mongo-ruby-driver raises an exception.

So, Mongo plugin resets buffer_chunk_limit if configurated value is larger than above limitation:

  • Before v1.8, max of buffer_chunk_limit is 2MB

  • After v1.8, max of buffer_chunk_limit is 8MB

Disable collection check

Mongo plugin checks a collection’s configuration to prevent unexpected insertion to existing collection. In tag mapped mode, Mongo plugin accesses many collections. In this case, collection checking is not usable because current configuration format cannot write multiple values in one parameter. So, if you disable this checking, put “‘disable_collection_check true“` in match.

<match forward.*>
  ...

  disable_collection_check true

  ...
</match>

Tool

You can tail mongo capped collection.

$ mongo-tail -f

Test

Run following command:

$ bundle exec rake test

You can use ‘mongod’ environment variable for specified mongod:

$ mongod=/path/to/mongod bundle exec rake test

Note that source code in test/tools are from mongo-ruby-driver.

TODO

More configuration

  • Multi process

  • etc

Copyright

Copyright

Copyright © 2011- Masahiro Nakagawa

License

Apache License, Version 2.0