Fluentd Blog

Fluentd v0.12.20 has been released

Hi users!

We have released Fluentd version 0.12.20. Here are the changes:

New features / Enhancement

• in_forward: Add skip_invalid_event parameter to check and skip invalid event: #766
• in_tail: Add multiline_flush_interval parameter for periodic flush with multiline format: #775
• filter_record_transformer: Improve ruby placeholder performance and adding record["key"] syntax: #766
• Add on_exit_process hook point to DetachProcessMixin: #757
• Add descriptions to BufferedOutput and TimeSlicedOutput: #759 and #760

Bug fixes

• parser: Don't use BigDecimal in JSON parsing result with Oj: #778
• config: Fix the regression of unused parameter warning inside sub section: #765

in_tail: multiline_flush_interval parameter

In multiline mode, format_firstline is a trigger for flushing buffered event. This has one limiation that next event is needed. So if next event is delayed, flushing event is also delayed.

v0.12.20 introduces multiline_flush_interval to resolve this problem. If set multiline_flush_interval 5s, in_tail flushes buffered event after 5 seconds from last emit.

filter_record_transformer: New record["key"] syntax

record_transformer filter uses ${key} to access key field of event record. This is confusing syntax because ${time} means event time, not time field of event record.

Since v0.12.20, record_transformer supports ${record["key"]} syntax. This is same syntax with record_modifier and elasticsearch_dynamic. We recomed to use ${record["key"]} instead of ${key}.

<record>
  foo ${foo}           # existing syntax
  foo ${record["foo"]} # new syntax
</record>

Lastly, v0.12.20 docker image has also been available on Docker Hub.

Happy logging!