Parse format mixed logs.
fluent-plugin-multi-format-parser | fluentd | ruby |
---|---|---|
>= 1.0.0 | >= v0.14.0 | >= 2.1 |
< 1.0.0 | >= v0.12.0 | >= 1.9 |
Use RubyGems:
fluent-gem install fluent-plugin-multi-format-parser
This plugin is a parser plugin. After installed, you can use multi_format
in <parse>
supported plugins.
Use multiple <pattern>
s to specify multiple parser formats.
<source>
@type udp
tag logs.multi
<parse>
@type multi_format
<pattern>
format apache
</pattern>
<pattern>
format json
time_key timestamp
</pattern>
<pattern>
format none
</pattern>
</parse>
</source>
multi_format
tries pattern matching from top to bottom and returns parsed result when matched.
Available format patterns and parameters are depends on Fluentd parsers. See parser plugin document for more details.
Put <pattern>
s inside <parse>
.
<filter app.**>
@type parser
key_name message
<parse>
@type multi_format
<pattern>
format json
</pattern>
<pattern>
format regexp
expression /...your regexp pattern.../
</pattern>
<pattern>
format none
</pattern>
</parse>
</filter>
Use format
instead of <parse></parse>
.
<filter app.**>
@type parser
key_name message
format multi_format
<pattern>
format json
</pattern>
<pattern>
format /...your regexp pattern.../
</pattern>
<pattern>
format none
</pattern>
</filter>
Sometimes it may be useful to know which pattern was used. Since pareser usage may not support retagging, there is an option to add a format name field and/or index field.
Example:
<filter app.**>
@type parser
key_name message
<parse>
@type multi_format
# if set, add this key to record with value being pattern format name
# (format_name key)
format_key 'format'
# If set, add index of the matching pattern as this key (0-based)
index_key 'parser_index'
<pattern>
format json
# set format name for this pattern. If unset, uses format name
# followed by index (in this case would be 'json#0')
format_name 'json'
</pattern>
<pattern>
format regexp
format_name 'MyRefex'
expression /...your regexp pattern.../
</pattern>
<pattern>
format none
format_name 'unparsed'
</pattern>
</parse>
</filter>
This plugin doesn't work with multiline
parsers because parser itself doesn't store previous lines.
Author | Masahiro Nakagawa |
Copyright | Copyright (c) 2014- Masahiro Nakagawa |
License | Apache License (2.0) |