Currently there is a flag in the QueryParseContext that keeps track
of whether an inner query sits inside a filter and should therefore
produce an unscored lucene query. This is done in the
parseInnerFilter...() methods that are called in the fromXContent()
methods or the parse() methods we haven't refactored yet. This needs
to move to the toQuery() method in the refactored builders, since the
query builders themselves have no information about the parent query
they might be nested in.
This PR moves the isFilter flag from the QueryParseContext to the re-
cently introduces QueryShardContext. The parseInnerFilter... methods
need to stay in the QueryParseContext for now, but they already delegate
to the flag that is kept in QueryShardContext. For refactored queries
(like BoolQueryBuilder) references to isFilter() are moved from the
parser to the corresponding builder. Builders where the inner query
was previously parsed using parseInnerFilter...() now use a newly
introduces toFilter(shardContext) method that produces the nested lucene
query with the filter context flag switched on.
Closes#12731
Due to the limited abilities of parsing of dynamic (not configured) arguments
like `http.cors.enabled`, that dont map to a command line argument but will
become configuration, we need to mention explicitely, that those dynamic arguments
must come last.
Also fixed some mentions of a memory index setting, that does not exist anymore.
Closes#12758
This creates a module in qa called vagrant that can be run if you have
vagrant and virtualbox installed and will run the packaging tests in trusty
and centos-7.0. You can ask it to run tests in other linuxes. This is the full
list:
* precise aka Ubuntu 12.04
* trusty aka Ubuntu 14.04
* vivid aka Ubuntun 15.04
* wheezy aka Debian 7, the current debian oldstable distribution
* jessie aka Debian 8, the current debina stable distribution
* centos-6
* centos-7
* fedora-22
* oel-7
There is lots of documentation on how to do this in the TESTING.asciidoc.
Closes#12611
When installed as a service with a DEB or RPM package, we should gently wait for elasticsearch to stop (flushing indices on closing can take some time) and never kill the process.
Closes#11248
Moved the license checker config into the parent pom, and overrede
the license dir/target-to-check in distributions/pom.
Disabled the license checker explicitly for projects which run integration
tests but have no licenses dir:
* core
* distribution
* qa
* plugins/delete-by-query
* plugins/mapper-size
* plugins/site-example
Closes#12752Closes#12754
This commit adds basic support to track the number of times scripts are
compiled and compiled scripts are evicted from the script cache. These
statistics are tracked at the node level.
Closes#12673
This commit updates the Zen Discovery documentation to explain which
nodes partcipate in master election (by default) as well as the
configuration parameters for controlling this.
Closes#12727
Bats testing uncovered a useless systemctl check, that resulted in an
error, because the systemctl file was uninstalled, but we hoped to
check for an explicetely configured SystemExitCode.
In addition we did not reload the systemctl configuration when uninstalling
elasticsearch, which now is fixed as well.
Closes#12682
this commit adds a simple integration test that starts a
node from a shaded jar, indexes a doc and retrieves it. It
also has some basic unittests that try to load shaded classes and ensure
that their counterpart is not in the classpath.
Closes#12711
Elasticsearch will create this if it doesn't exist if it cant but because
it doesn't own /etc/elasticsearch when installed by rpm and deb it can't
create /etc/elasticsearch/scripts.
Closes#12702
ClusterState has 3 different methods to access RoutingNodes:
* #routingNodes() - mutable version
* #getRoutingNodes() - delegates to #getReadOnlyRoutingNodes()
* #getReadOnlyRoutingNodes() - it's docs say `NOTE, the routing nodes are mutable, use them just for read operations`
The latter also reuses the instance that it creates. This has several problems beside the obvious:
* creating RoutingNodes is costly and should be done only if really needed ie. use cached version as much as possible
* the common case is ReadOnly but all kinds of things are called
* mutable version are only needed in one place and should only be used in the AllocationService
* RoutingNodes can freeze it's ShardRoutings but doesn't
* RoutingNodes should check if it's read-only or not
This commit fixed all the problems and special cases the mutable case such that all accesses via ClusterState#getRoutingNodes()
is read-only and RoutingNodes enforces this.
