Logstash 6.3.1 Release Notesedit

  • Adds a Persistent Queue repair utility, enabling self-recovery of corrupted PQs (Issue 9710).
  • Fixes two separate issues in Experimental Java Execution mode where complex pipeline configurations could fail to compile (Issue 9747, Issue 9745).
  • Fixes issue when running Logstash inside of a Docker container with Persistent Queue enabled where we incorrectly reported that there was insufficient space to allocate for the queue (Issue 9766).
  • Fixes issue in x-pack monitoring where queue_push_duration_in_millis was incorrectly reporting nanoseconds (Issue 9744).
  • Fixes an issue where Logstash could fail to start when its path.data is a symlink (Issue 9706).
  • Fixes issue with Netflow module where it could fail to populate [geoip_dst][autonomous_system] (Issue 9638).
  • Fixes a potential conflict with plugin depenencies that require Guava by explicitly loading Guava 22 (Issue 9592).


Netflow Codec

  • Fixes exception when receiving Netflow 9 from H3C devices.
  • Added support for Netflow 9 from H3C devices.
  • Fixes incorrect definitions of IE 231 and IE 232.
  • Fixes exceptions due to concurrent access of IPFIX templates.
  • Added support for Netflow 9 reduced-size encoding support.
  • Added support for Barracuda IPFIX Extended Uniflow.

Beats Input

  • Fixes an issue that prevented auto-recovery in certain failure modes.
  • Fixes an issue where trace-level logging omitted helpful context.

Kafka Input

  • Fix race-condition where shutting down a Kafka Input before it completes startup could cause Logstash to crash.
  • Upgrade Kafka client to version 1.1.0.

S3 Input

  • Avoid plugin crashes when encountering bad files in S3 buckets.
  • Log entry when bucket is empty.
  • Fixes additional_settings configuration option to properly symbolize keys for downstream library.

TCP Input

  • New configuration option to set TCP keep-alive.
  • Fixes an issue where the input could crash during shutdown, affecting pipeline reloads.

UDP Input

  • Mitigate memory leak in JRuby’s UDP implementation.

DNS Filter

  • Log timeouts as warn instead of error.
  • Allow concurrent queries when cache enabled.

Elasticsearch Filter

  • Fix: The filter now only calls filter_matched on events that actually matched. This fixes issues where all events would have success-related actions happened when no match had actually happened (add_tag, add_field, remove_tag, remove_field).

JDBC Static Filter

  • Fixed an issue where failing to specify index_columns would result in an obscure error message.

KV Filter

  • improves trim_key and trim_value to trim any sequence of matching characters from the beginning and ends of the corresponding keys and values; a previous implementation limited trim to a single character from each end, which was surprising.
  • fixes issue where we can fail to correctly break up a sequence that includes a partially-quoted value followed by another fully-quoted value by slightly reducing greediness of quoted-value captures.

Mutate Filter

  • Fix: when converting to float and float_eu, explicitly support same range of inputs as their integer counterparts.

Elasticsearch Output

  • Added support for customizing HTTP headers.
  • Log an error — not a warning — when ES raises an invalid_index_name_exception.
  • Improve plugin behavior when Elasticsearch is down on startup.

File Output

  • Fix a bug where flush interval was being called for each event when enabled

Kafka Output

  • Changed Kafka send errors to log as warn.
  • Upgrade Kafka client to version 1.1.0.

S3 Output

  • Fixes additional_settings configuration option to properly symbolize keys for downstream library.

SQS Output

  • Added the ability to send to a different account id’s queue.