webhdfs

The out_webhdfs TimeSliced Output plugin writes records into HDFS (Hadoop Distributed File System). By default, it creates files on an hourly basis. This means that when you first import records using the plugin, no file is created immediately. The file will be created when the time_slice_format condition has been met. To change the output frequency, please modify the time_slice_format value. This document doesn't describe all parameters. If you want to know full features, check the Further Reading section.

Install

out_webhdfs is included in td-agent by default (v1.1.10 or later). Fluentd gem users will have to install the fluent-plugin-webhdfs gem using the following command.

$ fluent-gem install fluent-plugin-webhdfs

HDFS Configuration

Append operations are not enabled by default on CDH. Please put these configurations into your hdfs-site.xml file and restart the whole cluster.

<property>
  <name>dfs.webhdfs.enabled</name>
  <value>true</value>
</property>

<property>
  <name>dfs.support.append</name>
  <value>true</value>
</property>

<property>
  <name>dfs.support.broken.append</name>
  <value>true</value>
</property>

Example Configuration

<match access.**>
  @type webhdfs
  host namenode.your.cluster.local
  port 50070
  path "/path/on/hdfs/access.log.%Y%m%d_%H.#{Socket.gethostname}.log"
  flush_interval 10s
</match>

Please see the Fluentd + HDFS: Instant Big Data Collection article for real-world use cases.

Please see the Config File article for the basic structure and syntax of the configuration file.

Parameters

type (required)

The value must be webhfds.

host (required)

The namenode hostname.

port (required)

The namenode port number.

path (required)

The path on HDFS. Please include "#{Socket.gethostname}" in your path to avoid writing into the same HDFS file from multiple Fluentd instances. This conflict could result in data loss.

Path value can contain time placeholders (see time_slice_format section). If path contains time placeholders, webhdfs output configures time_slice_format automatically with these placeholders.

Time Sliced Output Parameters (and overwritten values by out_webhdfs)

For advanced usage, you can tune Fluentd's internal buffering mechanism with these parameters.

time_slice_format

The time format used as part of the file name. The following characters are replaced with actual values when the file is created:

\%Y: year including the century (at least 4 digits)
\%m: month of the year (01..12)
\%d: Day of the month (01..31)
\%H: Hour of the day, 24-hour clock (00..23)
\%M: Minute of the hour (00..59)
\%S: Second of the minute (00..60)

The default format is %Y%m%d%H, which creates one file per hour. This parameter may be overwritten by path configuration.

time_slice_wait

The amount of time Fluentd will wait for old logs to arrive. This is used to account for delays in logs arriving to your Fluentd node. The default wait time is 10 minutes ('10m'), where Fluentd will wait until 10 minutes past the hour for any logs that occurred within the past hour.

For example, when splitting files on an hourly basis, a log recorded at 1:59 but arriving at the Fluentd node between 2:00 and 2:10 will be uploaded together with all the other logs from 1:00 to 1:59 in one transaction, avoiding extra overhead. Larger values can be set as needed.

buffer_type

The buffer type is memory by default (buf_memory). The file (buf_file) buffer type can be chosen as well. If you use file buffer type, buffer_path parameter is required.

buffer_queue_limit, buffer_chunk_limit

The length of the chunk queue and the size of each chunk, respectively. Please see the Buffer Plugin Overview article for the basic buffer structure. The default values are 64 and 8m, respectively. The suffixes "k" (KB), "m" (MB), and "g" (GB) can be used for buffer_chunk_limit.

flush_interval

The interval between data flushes. The default is unspecified, and buffer chunks will be flushed at the end of time slices. The suffixes "s" (seconds), "m" (minutes), and "h" (hours) can be used.

flush_at_shutdown

The boolean value to specify whether to flush buffer chunks at shutdown time, or not. The default is true. Specify true if you use memory buffer type.

retry_wait, max_retry_wait

The initial and maximum intervals between write retries. The default values are 1.0 and unset (no limit). The interval doubles (with +/-12.5% randomness) every retry until max_retry_wait is reached.

Since td-agent will retry 17 times before giving up by default (see the retry_limit parameter for details), the sleep interval can be up to approximately 131072 seconds (roughly 36 hours) in the default configurations.

retry_limit, disable_retry_limit

The limit on the number of retries before buffered data is discarded, and an option to disable that limit (if true, the value of retry_limit is ignored and there is no limit). The default values are 17 and false (not disabled). If the limit is reached, buffered data is discarded and the retry interval is reset to its initial value (retry_wait).

num_threads

The number of threads to flush the buffer. This option can be used to parallelize writes into the output(s) designated by the output plugin. The default is 1.

log_level option

The log_level option allows the user to set different levels of logging for each plugin. The supported log levels are: fatal, error, warn, info, debug, and trace.

Please see the logging article for further details.