05 Aug 2016
Fixed an issue which prevents java client from running explain statements.
Fixed an issue which caused
current_timestamp as part of a generated expression to not work correctly. This fix will only affect new tables. Tables which already contain a generated column which uses
current_timestamp will need to be recreated.
Improved error handling when grouping by invalid data types.
Fixed an issue when ordering by a function on a joined query.
Fixed an issue which could lead to wrong Error messages due to a race condition.
Fix: Arrays of geo_point can be used in insert/update statements as literals as well as bind arguments.
Fix: If a graceful shutdown of a node was aborted, sql statement processing on that node was still disabled.
Fixed a read isolation issue which could cause joins involving system and blob tables to return inconsistent results.
BREAKING: On Hunspell dictionary configuration, the parameter indices.analysis.hunspell.dictionary.location has been removed and
Fixed a bug that prevented join and sort on blob tables.
Adapted behaviour of stats.enabled, stats.jobs_log_size and stats.operations_log_size settings to match behaviour of other cluster wide settings like for example minimum_master_nodes.
Fix: Do not fetch table stats when node is not fully available yet. This could have led to NullPointerExceptions during long recovery process on node start.
Fixed an issue which prevented running join queries on empty partitioned tables.
hostname column of the
sys.nodes table returns a hostname instead of an IP address.
Fixed an issue that caused select statements to get stuck if used as part of a bulk request. Now a proper error is raised.
Updated Elasticsearch to 2.3.4
within scalar function now allows querying geo_shapes within geo_shapes.