From elasticsearch 0.90.6, when you have templates files defined in `config/templates` dir, rivers don't start anymore.
Steps to reproduce:
Create `config/templates/default.json`:
```javascript
{
default:
{
"template" : "*",
"mappings" : {
"_default_" : {
}
}
}
}
```
Start a dummy river:
```sh
curl -XPUT 'localhost:9200/_river/my_river/_meta' -d '{ "type" : "dummy" }'
```
It gives:
```
[2014-01-01 22:08:38,151][INFO ][cluster.metadata ] [Forge] [_river] creating index, cause [auto(index api)], shards [1]/[1], mappings [_default_]
[2014-01-01 22:08:38,239][INFO ][river.routing ] [Forge] no river _meta document found, retrying in 1000 ms
[2014-01-01 22:08:38,245][INFO ][cluster.metadata ] [Forge] [_river] update_mapping [my_river] (dynamic)
[2014-01-01 22:08:38,250][INFO ][river.routing ] [Forge] no river _meta document found, retrying in 1000 ms
[2014-01-01 22:08:39,244][INFO ][river.routing ] [Forge] no river _meta document found, retrying in 1000 ms
[2014-01-01 22:08:39,252][INFO ][river.routing ] [Forge] no river _meta document found, retrying in 1000 ms
[2014-01-01 22:08:40,246][INFO ][river.routing ] [Forge] no river _meta document found, retrying in 1000 ms
[2014-01-01 22:08:40,254][INFO ][river.routing ] [Forge] no river _meta document found, retrying in 1000 ms
[2014-01-01 22:08:41,246][INFO ][river.routing ] [Forge] no river _meta document found, retrying in 1000 ms
[2014-01-01 22:08:41,255][INFO ][river.routing ] [Forge] no river _meta document found, retrying in 1000 ms
[2014-01-01 22:08:42,249][WARN ][river.routing ] [Forge] no river _meta document found after 5 attempts
[2014-01-01 22:08:42,257][WARN ][river.routing ] [Forge] no river _meta document found after 5 attempts
```
With elasticsearch 0.90.2 or with no template file in `config/templates` dir, it gives:
```
[2014-01-01 22:22:32,096][INFO ][cluster.metadata ] [Forge] [_river] creating index, cause [auto(index api)], shards [1]/[1], mappings []
[2014-01-01 22:22:32,221][INFO ][cluster.metadata ] [Forge] [_river] update_mapping [my_river] (dynamic)
[2014-01-01 22:22:32,228][INFO ][river.dummy ] [Forge] [dummy][my_river] create
[2014-01-01 22:22:32,228][INFO ][river.dummy ] [Forge] [dummy][my_river] start
[2014-01-01 22:22:32,234][INFO ][cluster.metadata ] [Forge] [_river] update_mapping [my_river] (dynamic)
```
Closes#4577.
Closes#4656.
StringFieldMapper.toXContent uses the defaults for analyzed fields in order to
know which options to add to the builder. This means that if the field is not
analyzed and has norms enabled, it will omit to emit `norms.enabled: true`.
Parsing the mapping again will result in a StringFieldMapper that has norms
disabled.
The same fix applies to index options.
Close#4760
If elasticsearch was started in the foreground an immediate exit on startup
led to logging in the logfile, where as when starting in the background,
an immediate exit logged to stdout.
Closes#4805
Added base TransportAction class for master read operations that execute locally or not depending on the request class (local flag).
Added support for local flag where missing, in a backwards compatible manner:
- IndicesExistsRequest
- GetAliasesRequest (get alias api, aliases exist api)
- TypesExistsRequest
- GetIndexTemplatesRequest (get template, template exists)
- GetSettingsRequest
- GetRepositoriesRequest
- PendingClusterTasks
Added parsing of the local flag where missing in Rest*Action.
Updated SPEC adding local flag param where missing and added REST tests that contain use of the local flag where it was just added.
Closes#3345
* Make it clearer that `aggs` is an allowed synomym
for the `aggregations` key
* Fix broken example in for datehistogram, `1.5M` is
not an allowed interval
* Make use of colon before examples consistent
* Fix typos
Including:
* change some string params to list
* make some params or bodies required
* removed incorrect URLs
* removed incorrect params
* Fixed name of pending_tasks
During query parsing if a filter is encountered that extends from NoCacheFilter then the filter will not be given to the filter cache (also not wrapped in FilterCacheFilterWrapper).
Also if a filter directly or indirectly wraps a NoCacheFilter then that filter will also not be cached.
Relates to #4757
Although the empty paths are still registered in the java RestActions, that is only to return a meaningful error in case the index is not provided.
Index/indices are now mandatory.
Add transport.publish_port setting to allow users to specify the port
other cluster members should use when connecting to an instance. This
is needed for systems such as OpenShift, where cluster communication
needs to use a publicly accessibly proxy port, because the normal port
(9300) is bound to a private loopback IP address.
If a thread is not alive getting ThreadMXBean#getThreadInfo(long[], int)
places null elemnents in the returned array which are not repected
in the HotTheards API.
Closes#4775
