2013年09月11日

This Week in Elasticsearch - September 11, 2013

By Luca Cavanna

Welcome to This Week in Elasticsearch. In this roundup, we try to inform you about the latest and greatest changes in Elasticsearch. We cover what happened in the Github repositories, as well as many Elasticsearch events happening worldwide, and give you a small peek into the future of the project.

Elasticsearch core

  • The free disk space on the nodes can now influence the shard allocation using the brand new DiskThresholdDecider, which can be enabled through the cluster.routing.allocation.disk.threshold_enabled cluster setting (#3480, 0.90 & master)
  • Added the clear scroll API to clear all the resources associated with a specific scroll id (#3657, 0.90 & master)
  • Added the rest.action.multi.allow_explicit_index configuration option (defaults to true), which allows rejection of any multi-index request that contains the index name in the request body (#3636, 0.90 & master)
  • Added the processors configuration option to explicitly provide the number of processors available to an Elasticsearch node, which influences thread pools and network workers sizing (#3643, 0.90 & master)
  • Optimized the compression ratio of the field data for multi-valued fields in order to improve faceting performance  (#3623, 0.90 & master)
  • Deprecated the use of named filters in favor of named queries, following the work that was previously done on extending named filters to named queries (commit, 0.90)
  • Removed named filters in favor of named queries, again following the work that was previously done on extending named filters to named queries (#3644, master)
  • Improved the geo_shape filter so that it can handle multiple geo_shape fields (#3242, 0.90 & master)
  • Made sure that empty statistics are returned for the recently added completion feature. If a shard was not fully initialized, the completion suggester would throw an exception. It now gracefully returns empty statistics instead of throwing an exception (#3619, master & 0.90)
  • An exception is now thrown if reserved characters are used as input for the completion feature (#3648, 0.90 & master)
  • Solved potential resource leak in the statistics for the completion feature (#3652, 0.90 & master)
  • Reverted the work done on improving the fast vector highlighter support for complex queries (#3357) as it could influence the cluster performance and stability (commit, 0.90 & master)
  • Made sure LZF transport compression is not used by default on Solaris, as it might cause segmentation fault (#3634, 0.90 & master)
  • Fixed potential synchronization issue when Elasticsearch transport layer connects to and disconnects from network channels (commit, 0.90 & master)
  • Fixed a minor bug in the Java API that prevented from using the toString method of the SuggestResponse class (#3661, 0.90 & master)
  • The plugin script supports now removing a plugin providing its name, the same way it was installed in the first place (#3421, 0.90 & master)
  • The plugin script supports now the silent mode (#3628, 0.90 & master)
  • Progress has been made on moving the documentation to the main repository, so that we can easily maintain different versions of our reference per branch

Elasticsearch ecosystem

Here's some more information about what is happening in the ecosystem we are maintaining around Elasticsearch, including plugin and driver releases, as well as news about Logstash and Kibana.

  • Logstash version 1.2.1 was released. You can find more information in the changelog
  • The Elasticsearch puppet module version 0.1.3 has been released
  • The .NET client NEST has been released in version 0.11.4.0
  • The first version of the new web administration tool for Elasticsearch, called elasticsearch-kopf, has been released

Slides

Where to find Us

Australia

On September 19th, the Melbourne Search Users Group will have talk by RealEstate.com.au about their migration to Elasticsearch.

Czech Republic

Should you find yourself in Prague for WebExpo Prague 2013, stop in to see Karel Minarik and Honza Kral co-present Elasticsearch: Beyond Full Text Search.

Denmark

Alex Reelsen will be attending to GOTO Aarhus with a few other folks from the Elasticsearch team on September 30th - October 2nd. You can pop by our booth to meet us, or just head to Alex's talk Elasticsearch - Beyond Full Text Search.

France

Germany

Shay Banon and Leslie Hawthorn will be attending Monitorama on September 19-20th in Berlin. Ping either of them on Twitter if you want to talk all things Elasticsearch over a tasty beverage.

Norway

The fine folks spearheading the Oslo Elasticsearch Users Group will host their inaugural meetup on Thursday, September 26th. Martijn van Groningen will join the festivities to talk about new features in the works for Elasticsearch.

Ukraine

RubyShift 2013 kicks off on Friday, September 27th in Kiev. Karel Minarik will be there, giving a presentation on Elasticsearch: Search and Analytics the first day of the conference.

United States

Where to Find You

Are you hosting an Elasticsearch meetup or giving a talk about Elasticsearch? We would love to know so we can feature that information in future editions of This Week in Elasticsearch. Just let our Community Manager know what you're up to and we're happy to help promote your efforts.

Oh yeah, we're also hiring. If you'd like us to find you for employment purposes, just drop us a note.  We care more about your skill set and passion for Elasticearch, Kibana and Logstash than where you rest your head.

Training

If you are interested in Elasticsearch training we have courses taught by our core developers coming up in: