fluent-plugin-bigquery

Fluentd output plugin to load/insert data into Google BigQuery.

• Plugin type: BufferedOutput

• insert data over streaming inserts
  • for continuous real-time insertions
  • https://developers.google.com/bigquery/streaming-data-into-bigquery#usecases
• load data
  • for data loading as batch jobs, for big amount of data
  • https://developers.google.com/bigquery/loading-data-into-bigquery

Current version of this plugin supports Google API with Service Account Authentication, but does not support OAuth flow for installed applications.

Version Information

v1.0.0 or later supports fluentd-0.14.0 or later. If you use fluentd-0.12.x, please use v0.4.x.

I recommend to update fluentd version to v0.14.x or later.

With docker image

If you use official alpine based fluentd docker image (https://github.com/fluent/fluentd-docker-image), You need to install bigdecimal gem on your own dockerfile. Because alpine based image has only minimal ruby environment in order to reduce image size. And in most case, dependency to embedded gem is not written on gemspec. Because embbeded gem dependency sometimes restricts ruby environment.

Configuration

Options

name	type	required?	placeholder?	default	description
method	string	no	no	insert	insert (Streaming Insert) or load (load job)
auth_method	enum	yes	no	private_key	private_key or json_key or compute_engine or application_default
email	string	yes (private_key)	no	nil	GCP Service Account Email
private_key_path	string	yes (private_key)	no	nil	GCP Private Key file path
private_key_passphrase	string	yes (private_key)	no	nil	GCP Private Key Passphrase
json_key	string	yes (json_key)	no	nil	GCP JSON Key file path or JSON Key string
project	string	yes	yes	nil
dataset	string	yes	yes	nil
table	string	yes (either tables)	yes	nil
tables	array(string)	yes (either table)	yes	nil	can set multi table names splitted by ,
template_suffix	string	no	yes	nil	can use %{time_slice} placeholder replaced by time_slice_format
auto_create_table	bool	no	no	false	If true, creates table automatically
skip_invalid_rows	bool	no	no	false	Only insert method.
max_bad_records	integer	no	no	0	Only load method. If the number of bad records exceeds this value, an invalid error is returned in the job result.
ignore_unknown_values	bool	no	no	false	Accept rows that contain values that do not match the schema. The unknown values are ignored.
schema	array	yes (either fetch_schema or schema_path)	no	nil	Schema Definition. It is formatted by JSON.
schema_path	string	yes (either fetch_schema)	no	nil	Schema Definition file path. It is formatted by JSON.
fetch_schema	bool	yes (either schema_path)	no	false	If true, fetch table schema definition from Bigquery table automatically.
fetch_schema_table	string	no	yes	nil	If set, fetch table schema definition from this table, If fetch_schema is false, this param is ignored
schema_cache_expire	integer	no	no	600	Value is second. If current time is after expiration interval, re-fetch table schema definition.
replace_record_key	bool	no	no	false	see examples.
replace_record_key_regexp{1-10}	string	no	no	nil	see examples.
insert_id_field	string	no	no	nil	Use key as insert_id of Streaming Insert API parameter.
add_insert_timestamp	string	no	no	nil	Adds a timestamp column just before sending the rows to BigQuery, so that buffering time is not taken into account. Gives a field in BigQuery which represents the insert time of the row.
allow_retry_insert_errors	bool	no	false	Retry to insert rows when an insertErrors occurs. There is a possibility that rows are inserted in duplicate.
request_timeout_sec	integer	no	no	nil	Bigquery API response timeout
request_open_timeout_sec	integer	no	no	60	Bigquery API connection, and request timeout. If you send big data to Bigquery, set large value.
time_partitioning_type	enum	no (either day)	no	nil	Type of bigquery time partitioning feature(experimental feature on BigQuery).
time_partitioning_expiration	time	no	no	nil	Expiration milliseconds for bigquery time partitioning. (experimental feature on BigQuery)

Buffer section

name	type	required?	default	description
@type	string	no	memory (insert) or file (load)
chunk_limit_size	integer	no	1MB (insert) or 1GB (load)
total_limit_size	integer	no	1GB (insert) or 32GB (load)
chunk_records_limit	integer	no	500 (insert) or nil (load)
flush_mode	enum	no	interval	default, lazy, interval, immediate
flush_interval	float	no	0.25 (insert) or nil (load)
flush_thread_interval	float	no	0.05 (insert) or nil (load)
flush_thread_burst_interval	float	no	0.05 (insert) or nil (load)

And, other params (defined by base class) are available

see. https://github.com/fluent/fluentd/blob/master/lib/fluent/plugin/output.rb

Inject section

It is replacement of previous version time_field and time_format.

For example.

<inject>
  time_key time_field_name
  time_type string
  time_format %Y-%m-%d %H:%M:%S
</inject>

name	type	required?	default	description
hostname_key	string	no	nil
hostname	string	no	nil
tag_key	string	no	nil
time_key	string	no	nil
time_type	string	no	nil
time_format	string	no	nil
localtime	bool	no	true
utc	bool	no	false
timezone	string	no	nil

see. https://github.com/fluent/fluentd/blob/master/lib/fluent/plugin_helper/inject.rb

Examples

Streaming inserts

Configure insert specifications with target table schema, with your credentials. This is minimum configurations:

<match dummy>
  @type bigquery

  method insert    # default

  auth_method private_key   # default
  email xxxxxxxxxxxx-xxxxxxxxxxxxxxxxxxxxxx@developer.gserviceaccount.com
  private_key_path /home/username/.keys/00000000000000000000000000000000-privatekey.p12
  # private_key_passphrase notasecret # default

  project yourproject_id
  dataset yourdataset_id
  table   tablename

  schema [
    {"name": "time", "type": "INTEGER"},
    {"name": "status", "type": "INTEGER"},
    {"name": "bytes", "type": "INTEGER"},
    {"name": "vhost", "type": "STRING"},
    {"name": "path", "type": "STRING"},
    {"name": "method", "type": "STRING"},
    {"name": "protocol", "type": "STRING"},
    {"name": "agent", "type": "STRING"},
    {"name": "referer", "type": "STRING"},
    {"name": "remote", "type": "RECORD", "fields": [
      {"name": "host", "type": "STRING"},
      {"name": "ip", "type": "STRING"},
      {"name": "user", "type": "STRING"}
    ]},
    {"name": "requesttime", "type": "FLOAT"},
    {"name": "bot_access", "type": "BOOLEAN"},
    {"name": "loginsession", "type": "BOOLEAN"}
  ]
</match>

For high rate inserts over streaming inserts, you should specify flush intervals and buffer chunk options:

<match dummy>
  @type bigquery

  method insert    # default
  
  <buffer>
    flush_interval 0.1  # flush as frequent as possible
    
    buffer_queue_limit 10240        # 1MB * 10240 -> 10GB!
    
    flush_thread_count 16
  </buffer>
  
  auth_method private_key   # default
  email xxxxxxxxxxxx-xxxxxxxxxxxxxxxxxxxxxx@developer.gserviceaccount.com
  private_key_path /home/username/.keys/00000000000000000000000000000000-privatekey.p12
  # private_key_passphrase notasecret # default

  project yourproject_id
  dataset yourdataset_id
  tables  accesslog1,accesslog2,accesslog3

  schema [
    {"name": "time", "type": "INTEGER"},
    {"name": "status", "type": "INTEGER"},
    {"name": "bytes", "type": "INTEGER"},
    {"name": "vhost", "type": "STRING"},
    {"name": "path", "type": "STRING"},
    {"name": "method", "type": "STRING"},
    {"name": "protocol", "type": "STRING"},
    {"name": "agent", "type": "STRING"},
    {"name": "referer", "type": "STRING"},
    {"name": "remote", "type": "RECORD", "fields": [
      {"name": "host", "type": "STRING"},
      {"name": "ip", "type": "STRING"},
      {"name": "user", "type": "STRING"}
    ]},
    {"name": "requesttime", "type": "FLOAT"},
    {"name": "bot_access", "type": "BOOLEAN"},
    {"name": "loginsession", "type": "BOOLEAN"}
  ]
</match>

Important options for high rate events are:

• tables
  • 2 or more tables are available with ',' separator
  • out_bigquery uses these tables for Table Sharding inserts
  • these must have same schema
• buffer/chunk_limit_size
  • max size of an insert or chunk (default 1000000 or 1MB)
  • the max size is limited to 1MB on BigQuery
• buffer/chunk_records_limit
  • number of records over streaming inserts API call is limited as 500, per insert or chunk
  • out_bigquery flushes buffer with 500 records for 1 inserts API call
• buffer/queue_length_limit
  • BigQuery streaming inserts needs very small buffer chunks
  • for high-rate events, buffer_queue_limit should be configured with big number
  • Max 1GB memory may be used under network problem in default configuration
    • chunk_limit_size (default 1MB) x queue_length_limit (default 1024)
• buffer/flush_thread_count
  • threads for insert api calls in parallel
  • specify this option for 100 or more records per seconds
  • 10 or more threads seems good for inserts over internet
  • less threads may be good for Google Compute Engine instances (with low latency for BigQuery)
• buffer/flush_interval
  • interval between data flushes (default 0.25)
  • you can set subsecond values such as 0.15 on Fluentd v0.10.42 or later

See Quota policy section in the Google BigQuery document.

Load

<match bigquery>
  @type bigquery

  method load

  <buffer>
  @type file
  path bigquery.*.buffer
  flush_interval 1800
  flush_at_shutdown true
  timekey_use_utc
  </buffer>

  auth_method json_key
  json_key json_key_path.json

  project yourproject_id
  dataset yourdataset_id
  auto_create_table true
  table yourtable%{time_slice}
  schema_path bq_schema.json
</match>

I recommend to use file buffer and long flush interval.

Authentication

There are four methods supported to fetch access token for the service account.

1. Public-Private key pair of GCP(Google Cloud Platform)'s service account
2. JSON key of GCP(Google Cloud Platform)'s service account
3. Predefined access token (Compute Engine only)
4. Google application default credentials (http://goo.gl/IUuyuX)

Public-Private key pair of GCP's service account

The examples above use the first one. You first need to create a service account (client ID), download its private key and deploy the key with fluentd.

JSON key of GCP(Google Cloud Platform)'s service account

You first need to create a service account (client ID), download its JSON key and deploy the key with fluentd.

<match dummy>
  @type bigquery

  auth_method json_key
  json_key /home/username/.keys/00000000000000000000000000000000-jsonkey.json

  project yourproject_id
  dataset yourdataset_id
  table   tablename
  ...
</match>

You can also provide json_key as embedded JSON string like this. You need to only include private_key and client_email key from JSON key file.

<match dummy>
  @type bigquery

  auth_method json_key
  json_key {"private_key": "-----BEGIN PRIVATE KEY-----\n...", "client_email": "xxx@developer.gserviceaccount.com"}

  project yourproject_id
  dataset yourdataset_id
  table   tablename
  ...
</match>

Predefined access token (Compute Engine only)

When you run fluentd on Googlce Compute Engine instance, you don't need to explicitly create a service account for fluentd. In this authentication method, you need to add the API scope "https://www.googleapis.com/auth/bigquery" to the scope list of your Compute Engine instance, then you can configure fluentd like this.

<match dummy>
  @type bigquery

  auth_method compute_engine

  project yourproject_id
  dataset yourdataset_id
  table   tablename

  ...
</match>

Application default credentials

The Application Default Credentials provide a simple way to get authorization credentials for use in calling Google APIs, which are described in detail at http://goo.gl/IUuyuX.

In this authentication method, the credentials returned are determined by the environment the code is running in. Conditions are checked in the following order:credentials are get from following order.

1. The environment variable GOOGLE_APPLICATION_CREDENTIALS is checked. If this variable is specified it should point to a JSON key file that defines the credentials.
2. The environment variable GOOGLE_PRIVATE_KEY and GOOGLE_CLIENT_EMAIL are checked. If this variables are specified GOOGLE_PRIVATE_KEY should point to private_key, GOOGLE_CLIENT_EMAIL should point to client_email in a JSON key.
3. Well known path is checked. If file is exists, the file used as a JSON key file. This path is $HOME/.config/gcloud/application_default_credentials.json.
4. System default path is checked. If file is exists, the file used as a JSON key file. This path is /etc/google/auth/application_default_credentials.json.
5. If you are running in Google Compute Engine production, the built-in service account associated with the virtual machine instance will be used.
6. If none of these conditions is true, an error will occur.

Table id formatting

this plugin supports fluentd-0.14 style placeholder.

strftime formatting

table and tables options accept Time#strftime format to construct table ids. Table ids are formatted at runtime using the chunk key time.

see. http://docs.fluentd.org/v0.14/articles/output-plugin-overview

For example, with the configuration below, data is inserted into tables accesslog_2014_08, accesslog_2014_09 and so on.

<match dummy>
  @type bigquery

  ...

  project yourproject_id
  dataset yourdataset_id
  table   accesslog_%Y_%m

  <buffer time>
    timekey 1d
  </buffer>
  ...
</match>

record attribute formatting

The format can be suffixed with attribute name.

CAUTION: format is different with previous version

<match dummy>
  ...
  table   accesslog_${status_code}

  <buffer status_code>
  </buffer>
  ...
</match>

If attribute name is given, the time to be used for formatting is value of each row. The value for the time should be a UNIX time.

time_slice_key formatting

Instead, Use strftime formatting.

strftime formatting of current version is based on chunk key. That is same with previous time_slice_key formatting .

Date partitioned table support

this plugin can insert (load) into date partitioned table.

Use placeholder.

<match dummy>
  @type bigquery

  ...
  table   accesslog$%Y%m%d

  <buffer time>
    timekey 1d
  </buffer>
  ...
</match>

But, Dynamic table creating doesn't support date partitioned table yet.

Dynamic table creating

When auto_create_table is set to true, try to create the table using BigQuery API when insertion failed with code=404 "Not Found: Table ...". Next retry of insertion is expected to be success.

NOTE: auto_create_table option cannot be used with fetch_schema. You should create the table on ahead to use fetch_schema.

<match dummy>
  @type bigquery

  ...

  auto_create_table true
  table accesslog_%Y_%m

  ...
</match>

Table schema

There are three methods to describe the schema of the target table.

1. List fields in fluent.conf
2. Load a schema file in JSON.
3. Fetch a schema using BigQuery API

The examples above use the first method. In this method, you can also specify nested fields by prefixing their belonging record fields.

<match dummy>
  @type bigquery

  ...

  schema [
    {"name": "time", "type": "INTEGER"},
    {"name": "status", "type": "INTEGER"},
    {"name": "bytes", "type": "INTEGER"},
    {"name": "vhost", "type": "STRING"},
    {"name": "path", "type": "STRING"},
    {"name": "method", "type": "STRING"},
    {"name": "protocol", "type": "STRING"},
    {"name": "agent", "type": "STRING"},
    {"name": "referer", "type": "STRING"},
    {"name": "remote", "type": "RECORD", "fields": [
      {"name": "host", "type": "STRING"},
      {"name": "ip", "type": "STRING"},
      {"name": "user", "type": "STRING"}
    ]},
    {"name": "requesttime", "type": "FLOAT"},
    {"name": "bot_access", "type": "BOOLEAN"},
    {"name": "loginsession", "type": "BOOLEAN"}
  ]
</match>

This schema accepts structured JSON data like:

{
  "request":{
    "time":1391748126.7000976,
    "vhost":"www.example.com",
    "path":"/",
    "method":"GET",
    "protocol":"HTTP/1.1",
    "agent":"HotJava",
    "bot_access":false
  },
  "remote":{ "ip": "192.0.2.1" },
  "response":{
    "status":200,
    "bytes":1024
  }
}

The second method is to specify a path to a BigQuery schema file instead of listing fields. In this case, your fluent.conf looks like:

<match dummy>
  @type bigquery

  ...
  
  schema_path /path/to/httpd.schema
</match>

where /path/to/httpd.schema is a path to the JSON-encoded schema file which you used for creating the table on BigQuery. By using external schema file you are able to write full schema that does support NULLABLE/REQUIRED/REPEATED, this feature is really useful and adds full flexbility.

The third method is to set fetch_schema to true to enable fetch a schema using BigQuery API. In this case, your fluent.conf looks like:

<match dummy>
  @type bigquery

  ...
  
  fetch_schema true
  # fetch_schema_table other_table # if you want to fetch schema from other table
</match>

If you specify multiple tables in configuration file, plugin get all schema data from BigQuery and merge it.

NOTE: Since JSON does not define how to encode data of TIMESTAMP type, you are still recommended to specify JSON types for TIMESTAMP fields as "time" field does in the example, if you use second or third method.

Specifying insertId property

BigQuery uses insertId property to detect duplicate insertion requests (see data consistency in Google BigQuery documents). You can set insert_id_field option to specify the field to use as insertId property.

<match dummy>
  @type bigquery

  ...

  insert_id_field uuid
  schema [{"name": "uuid", "type": "STRING"}]
</match>

TODO

• OAuth installed application credentials support
• Google API discovery expiration
• check row size limits

Authors

• @tagomoris: First author, original version
• KAIZEN platform Inc.: Maintener, Since 2014.08.19
• @joker1007