Breaking changes:

• Java client: FilterBuilder and QueryBuilder throw ElasticSearchIllegalArgumentException #4199

New features:

• Snapshot/Restore API - Phase 1 #3826
• Aggregation Module - Phase 1 #3300
• Aggregations: Support in the Percolate API #4245

Enhancements:

• Lucene: Upgrade to Lucene 4.6.0 #4241
• Aggegrations: Terms aggregator supports filtering terms by a regex #4267
• Aggregations: Support for shard_size to terms aggregations #4242
• Open/close index API supports new acknowledgement mechanism #4169
• Indices aliases API uses new acknowledgement mechanism #4114
• PUT Mapping API uses new acknowledgement mechanism #4228
• PUT Mapping API: A date format can now be updated #3727
• Bulk API supports shard timeout like the index API #4220
• Query: A field data based TermsFilter has been added #4209
• Script fields: Native scripts support primitive arrays as values #4175
• Transport client: TransportClient.connectedNodes() contains latest node info when using SimpleNodeSampler #4162
• Scroll search: The SearchContext object could allocate an oversized array #4156
• Cat API: Add support for pending tasks #4251
• Cat API: Remove heap used from _cat/nodes and added RAM percentage #4151
• Cat API: Add h parameter to apis, allowing to return columns and descriptions #4262
• Nodes info: Added mlockall setting to process info output #4233
• Nodes JVM stats: Expose maximum heap settings and heap usage in percent of JVM memory #4145
• Index Stats: Add support for segments stats #4101
• Lucene: mmapfs is now the default store for 64-bit Linux in order to adhere with Lucene defaults #4134
• Postings highlighter now supports the highlight_query functionality #4121
• Deprecate numeric_range filter, add a field data execution option to range filter #4034
• Maven plugin updates
  • RPM plugin to 2.1-alpha-3 #4282
  • Jar plugin to 2.4 #4281
  • Resources plugin to 2.6 #4280
  • Compiler plugin to 3.1 #4279
  • Assembly plugin to 2.4 #4278
  • Eclipse plugin to 2.9 #4277
  • Source plugin to 2.2.1 #4276
  • Surefire plugin to 2.16 #4275
  • Dependency plugin to 2.8 #4274
  • Shade plugin to 2.2, supports to shade test artifact as well #4266

Bug fixes:

• Query: has_child filter and query could yield inconsistent results #4306
• Query: has_parent query could yield inconsistent results #4313
• Query: minScore was not inclusive #4303
• Query: AllTermQuery explanations were not accurate #4298
• Query: has_child query with score_mode=avg can emit infinity as score #4291
• Query: Fetch/Count could fail if executed on a relocated shard. #4273
• Query: has_child query could result in NullPointerException if number of results exceed certain limit #4135
• Query: Wrong result on bool filter with must_not and geo_distance #4130
• Query: Indices filter did not support _name parameter #4166
• Query: Indices filter parsed for indices to which it should not apply #2416
• Mapping: _default_ mapping not applied when using separate master/data nodes #4124
• Mapping: External method to set rootTypeParsers in DocumentMapperParser was incorrect #4113
• Highlighting: Hit object cache gets reset per hit #4106
• Highlighting: Postings highlighter doesn't return snippets when using "path":"just_name" #4116
• Rivers did not start correctly #4089
• Rivers: River instance not created after successful creation of _meta document #3840
• Cat API: _cat/indices waited for cluster health timeout on missing index #4296
• Fielddata: Ensuring using correct field data format #4270
• Startup: Prefer local primary shards to relocating primary shards during startup #4237
• Shard startup: IndexShardGatewayService should not call post_recovery if shard is in STARTED state #4147
• Index: BulkProcessor: Fixing a possible deadlock #4153
• Index templates: Loading templates via templates/ directory has been fixed #4235
• Robin engine: Fixed a NullPointerException in RobinEngine.acquireSearcher() if engine was already closed#4232
• Translog: Fixed a NullPointerException in FsTranslog when reverting transient log #4223
• Nodes stats: Running the node stats API, while a shard is moving onto the node, logs an exception #4203
• Warmer API: Creating an empty warmup query could lead to exceptions #4196
• PluginsService: Preventing a possible NullPointerException when starting elasticsearch with a different user #4186
• Windows startup: Using JAVA_OPTS could fail in service.bat #4127
• Aliases: Unable to create a nested filtered alias on a dataless master #4112
• Multi term vector request never returns if the shards for all requested documents are non existent #3989
• Maven: OOM when building with java6 #4293