TransportShardReplicationOperationAction is a mouthful and is the only thing we mean when we say replication. This commit also changes some related friends.
#10032 introduced the notion of sealing an index by marking it with a special read only marker, allowing for a couple of optimization to happen. The most important one was to speed up recoveries of shards where we know nothing has changed since they were online by skipping the file based sync phase. During the implementation we came up with a light notion which achieves the same recovery benefits but without the read only aspects which we dubbed synced flush. The fact that it was light weight and didn't put the index in read only mode, allowed us to do it automatically in the background which has great advantage. However we also felt the need to allow users to manually trigger this operation.
The implementation at #11179 added the sync flush internal logic and the manual (rest) rest API. The name of the API was modeled after the sealing terminology which may end up being confusing. This commit changes the API name to match the internal synced flush naming, namely `{index}/_flush/synced'.
On top of that it contains a couple other changes:
- Remove all java client API. This feature is not supposed to be called programtically by applications but rather by admins.
- Improve rest responses making structure similar to other (flush) API
- Change IndexShard#getOperationsCount to exclude the internal +1 on open shard . it's confusing to get 1 while there are actually no ongoing operations
- Some minor other clean ups
The bin/plugin script now uses the default CONF_DIR & CONF_FILE environment vars. This allows to install a plugin even if Elasticsearch has been installed with a RPM or a DEB package. This commit also adds testing files for TAR archive and plugins installation.
Closes#10673
Today, when loading plugins from the classpath we take the enumeration
given to us by the classloader and attempt to load every URL. This can
cause issues as certain classloaders, such as groovy's, will return the same
URL multiple times in the enumeration. When this happens, startup can fail
with guice errors as bindings have already been registered.
To workaround this, we create a set from the URLs returned by the classloader
to provide uniqueness.
We had some code that created a TermQuery instead of a TermsQuery when there
was a single term, which is not useful anymore now that TermsQuery rewrites to
a disjunction when there are few terms:
https://issues.apache.org/jira/browse/LUCENE-6360
Simplification of MultiValueMode by removing the apply and reduce
methods for each mode. This creates a more consistent environment for
sorting methods since all sorting must now go through select methods.
This allows for better error handling and better encapsulation for
sorting fields with multiple values.
Note that apply and reduce had inconsistencies in the code base
prior to this change since different calls were assuming that the
accumulator for apply was the first input versus the second input.
Also added is an UnsortedNumericDoubleValues interface to allow
customized values to be input into the different sort modes. This
prevents the need for apply/reduce outside of MultiValueMode.
closes#11290
In order to get some information if the TTL purger thread could
successfully delete all documents per bulk exection, this commit
adds some logging. TRACE level logging will potentially contain
a lot of information about all the bulk failures.
Closes#11019
Michael McCandless
David Pilato
Britta Weber
Igor Motov
Simon Willnauer
Boaz Leskes
Alex Chan
Tanguy Leroux
Eduardo Gurgel
Clinton Gormley
Robert Muir
Robert Muir
Ryan Ernst
jaymode
Adrien Grand
Jack Conradson
Shay Banon
Alexander Reelsen