…eck if doc exists or not, closes elastic#2047.

…ress.stored

…#1921.

This effectively happens because we don't flush before deleting a mapping, which causes it to still reply the changes happening for it on the transaction log, meaning the mapping are still around on the index level.

…trol back

…c#2051.

though harmless, since we don't have any "default" component, and effectively has not effect on the settings used, it still puts settings that we don't really need

The ability to use es.default prefix broke the deb package that uses it now.

…oses elastic#2052.

…failure, index compression works), closes elastic#2055.

…n it, which means one can't highlight on it when searching on _all, closes elastic#2062.

… request

Allow the use of "doc" as the update source when a script is not
specified.  New fields are added, existing fields are overwritten, and
maps are merged recursively.

(note, this will break plugins that implement a rest handler..., like wares, thrift, and memcached, until they are upgraded)

…ngling index, closes elastic#2065.

…JSON, closes elastic#2064.

…e), closes elastic#2071.

…c#2074.

…d transport stream

…no compressor found on transport layer, closes elastic#2076.

…ore eventually delete it based on its logic of when to delete a shard

…hem, closes elastic#2077.

…s dangling, harden when we delete the _state of an index

.

…ero it out so ti can be GC'ed

…elastic#2085.

…plain the score

… shards

…(linked/array), defaults to array, closes elastic#2086.

First phase at improving buffer management and reducing even further buffer copies. Introduce a BytesReference abstraction, allowing to more easily slice and "read/write references" from streams. This is the foundation for later using it to create smarter buffers on top of composite netty channels for example (which http now produces) as well as reducing buffer copies when sending transport/rest responses.

this allows for better buffer usage, specifically when forwarding requests to other nodes

with the new bytes reference abstraction, and the new composite buffer handling in frame decoder, it makes more sense to move back and use netty frame decoder again, since we can easily slice refernces acorss composite buffers, and reduce copies when using compression

…I if used), closes elastic#2092.

…its library if its not found

…trings, apply to HighlightedField (breaks backward for Java API from String to Text), closes elastic#2093.

By introducing the Text abstraction, we can keep (long) text fields in their UTF8 bytes format, and no need to convert them to a string when serializing it back to Json for example.

The first place we can apply this is to highlighted text, which can be long.. . This does breaks backward comp. for people using the Java API where the HighlightField now has a Text as its content, and not String.

…n when state persistence is disabled, so they can act on it if needed

…ic#2098.

…string we get

…ildcard field support, it no longer does

…ectly the utf8 bytes

…d above), closes elastic#2122.

…e used directly when sending it over with netty

…mprove netty receive buffer predictor, closes elastic#2124.

… fail, just use the default value (which people really never change).

…en receive buffer size predictor size based on it

…dd more options as env vars when using maven

…Factory

… as well, reformat a bit

…impler to know from stack traces the es version

…ic#2121

…ssue elastic#2120

…2138.

…reads (time out of interval)

… of using shard id

this allows to handle better cases where we need to cancel an existing recovery

Fixed type.

…lastic#2141

…) constructor, instead of just the default

…ts), closes elastic#2150

…reused

now that we support "null" queries, for example, when parsing a query and its lenient, make sure to handle it where applicable

…l to use the routing, closes elastic#2155.

…ions exposed via mapping API

…he removal of mappings

…f positions are omitted

…method

…ssible change in mappings, not just additions

…d optimize when its not, closes elastic#2189.

…#2192.

…e double check on the facet filter

…ough its disabled

…will also help in tests when testing plugin related functionality

…he optimizeQuery method

Streamline the use of minimum should match to all relevant queries to accept `minimum_should_match`, and allow all relevant queries to accept the advanced "string" based config

…a node

…the load when doing recovery, closes elastic#2199.

the index.shard.recovery.concurrent_streams controls how many streams are opened from a recovery source to a recovery target to transfer index files. Reduce it from 5 to 3 to reduce the load when doing recovery (for example, due to relocation).

Note, recent changes in network buffering will mean that recovery will progress considerably faster, so this change will not affect recovery times.

A Bulk UDP service is a service listening over UDP for bulk format requests. The idea is to provide a low latency UDP service that allows to easily index data that is not of critical nature.

The Bulk UDP service is disabled by default, but can be enabled by setting `bulk.udp.enabled` to `true`.

The bulk UDP service performs intenral bulk aggregation of the data and then flushes it based on several parametres:

* `bulk.udp.bulk_actions`: The number of actions to flush a bulk after, defaults to `1000`.
* `bulk.udp.bulk_size`: The size of the current bulk request to flush the request once exceeded, defaults to `5mb`.
* `bulk.udp.flush_interval`: An interval after which the current request is flushed, regarldess of the above limits. Defaults to `5s`.
* `bulk.udp.concurrent_requests`: The number on max in flight bulk requests allowed. Defaults to `4`.

The network settings allowed are:

* `bulk.udp.host`: The host to bind to, defualts to `network.host` which defaults to any.
* `bulk.udp.port`: The port to use, defaults to `9700-9800`.

Here is an example of how it can be used:

    > cat bulk.txt
    { "index" : { "_index" : "test", "_type" : "type1" } }
    { "field1" : "value1" }
    { "index" : { "_index" : "test", "_type" : "type1" } }
    { "field1" : "value1" }

    > cat bulk.txt | nc -w 0 -u localhost 9700

* Fixed an issue where dynamic update to minimum_master_nodes settings would not take immediate effect
* Added LocalNodeMasterListener support to the ClusterService. Enables listening to when the local node becomes/stopped being a master