Release 344 (9 Oct 2020)#
General#
Add
from_unixtime_nanos()
function. (#5046)Improve performance and latency of queries leveraging dynamic filters. (#5081, #5340)
Add
dynamic-filtering.service-thread-count
config property to specify number of threads used for processing dynamic filters on coordinator. (#5341)Extend Секреты в файлах конфигурации environment variable substitution to allow multiple replacements in a single configuration property. (#4345)
Remove the
fast-inequality-joins
configuration property. This feature is always enabled. (#5375)Use
timestamp(3) with time zone
rather thantimestamp(3)
for thequeries
,transactions
, andtasks
tables insystem.runtime
. (#5464)Improve performance and accuracy of
approx_percentile()
. (#5158)Improve performance of certain cross join queries. (#5276)
Prevent potential query deadlock when query runs out of memory. (#5289)
Fix failure due to rounding error when casting between two
timestamp
types with precision higher than 6. (#5310)Fix failure due to rounding error when casting between two
timestamp with time zone
types with precision higher than 3. (#5371)Fix column pruning for
EXPLAIN ANALYZE
. (#4760)Fix incorrect timestamp values returned by the
queries
,transactions
, andtasks
tables insystem.runtime
. (#5462)
Security#
Предупреждение
The file-based system and catalog access controls have changed in ways that reduce or increase permissions. Please, read these release notes carefully.
Change file-based catalog access control from deny to allow when table, schema, or session property rules are not defined. (#5039)
Add missing table rule checks for table and view DDL in file-based system access control. (#5039)
Add missing schema rule checks for create schema in file-based system access control. (#5039)
Add session property rules to file-based system access control. (#5039)
Add catalog regex to table and schema rules in file-based system access control. (#5039)
Change create, rename, alter, and drop table in file-based system controls to only check for table ownership. (#5039)
Change file-based system access control to support files without catalog rules defined. In this case, all access to catalogs is allowed. (#5039)
Change file-based system and catalog access controls to only show catalogs, schemas, and tables a user has permissions on. (#5039)
Change file-based catalog access control to deny permissions inspection and manipulation. (#5039)
Add file-based group provider. (#5028)
Hive connector#
Add support for
hive.security=allow-all
, which allows to skip all authorization checks. (#5416)Support Kerberos authentication for Hudi tables. (#5472)
Allow hiding Delta Lake tables from table listings such as
SHOW TABLES
orinformation_schema.tables
, as these tables cannot be queried by the Hive connector. This be enabled using thehive.hide-delta-lake-tables
configuration property. (#5430)Improve query concurrency by listing data files more efficiently. (#5260)
Fix Parquet encoding for timestamps before 1970-01-01. (#5364)
Kafka connector#
Expose message timestamp via
_timestamp
internal column. (#4805)Add predicate pushdown for
_timestamp
,_partition_offset
and_partition_id
columns. (#4805)
Phoenix connector#
Fix query failure when a column name in
CREATE TABLE
requires quoting. (#3601)
PostgreSQL connector#
Add support for setting a column comment. (#5307)
Add support for variable-precision
time
type. (#5342)Allow
CREATE TABLE
andCREATE TABLE AS
withtimestamp
andtimestamp with time zone
with precision higher than 6. The resulting column will be declared with precision of 6, maximal supported by PostgreSQL. (#5342)
SQL Server connector#
Improve performance of queries with aggregations and
WHERE
clause. (#5327)