Follow

Kinetica 7.0.3.1 Release Notes

Build version: 7.0.3.1

FIXES: 

KDEVOPS-451 Fix to correctly grab python api version
KDEVOPS-451 Move sub-builds after build/install dir cleaning
KDEVOPS-451 Update to use --copy-install-to on python API build
KDEVOPS-454 Added --copy-install-to for 'parallel' sub-builds
KDEVOPS-460 Remove MAC ODBC client from dist
KDEVOPS-460 Remove Mac ODBC client from make-dist.sh
KDEVOPS-464 Fix node.js build issue on Ubuntu
KDEVOPS-464 Reinstate (and correct) fix for node.js build where path is too long
KDEVOPS-464 Reverted node.js 8 commits - build was not reliable
KDEVOPS-464 Upgrade node.js to 8.x
KDEVOPS-467 Add path for cairo to core/lib to find libcairo.so.2
KDEVOPS-467 Fix issue where shebangs are reverted. Issue derived from re-installing set-upTools
KDEVOPS-467 Updated GaiaDevLibsVersion.h to force rebuild of pycairo
KDEVOPS-473 Updated Java dependency to 1.8 for KIO/spark
KGEO-506 Changed some Helper function classes to namespace
KGEO-506 Refactored GeometryUtils.cpp to multiple .cpp files and multiple Helper classes
KGEO-506 Removed Inline keywords for helper functions. Typedefs are now in namespaces
KGEO-586 Added baseline images for isochrones
KGEO-586 Added Isochrones endpoint - Modified contour endpoint to handle isochrone line - Added Isochrones kernel to get convex then concave hulls from isolines
KGEO-586 Fixed isochrones output WKT table and changed warning to UERR when min_level>max_level
KGEO-587

Added the QUERY_TARGET_NODE_LABEL option to the queries so that no more implicit restrictions required to specify target nodes indirectly - Revised the tests and options (removed blocked_nodes option; redundant)

KGEO-597 Added single/double precision rendering control to gpudb.conf - update conf.ini
KGEO-597 Added single/double precision rendering control to gpudb.conf
KGEO-600 Added lock files when importing data for graph tests - Fixed crash in isochrones - Minor changes to minimize testing time for graph and isochrones tests
KGEO-606 Use char64 instead of strings when passing constants to the graph server - Fixed isochrone test
KGEO-608 Fixed bug for ST_Concavehull 
KINFRA-1071 Add gpudb_job_canceller.py to automate job cancelling testing
KINFRA-1071 gpudb_job_canceller.py --continous arg is now -d duration and exit with ERR if jobs aren't canceled
KINFRA-1095 Add tbb and zmq valgrind suppressions  and update gpudb_cluster_vagrind to use them if found
KINFRA-1095 CMake copies scripts/gpudb_cluster_valgrind to the bin dir
KINFRA-1095 gpudb_cluster_vagrind added an 's' option for --suppression
KINFRA-1095 PGroupBySink.cpp don't do work on empty vector in synthesize_count_distinct_result
KINFRA-1095 scripts/gpudb_cluster_valgrind fix finding intel exe
KINFRA-1107 Identify deleted records from indexeded files not containing stencils
KINFRA-1120 If a newly added rank from /admin/add/ranks fails to sync types and tables in time, it shuts down and removes itself from the cluster
KINFRA-1120 Rebalance workers store shared pointers to GaiaSet in contexts instead of reference
KINFRA-1120 Worker_rebalance_sharded and Worker_rebalance_unsharded now set max concurrency to # of calc threads
KINFRA-1124 WorkItemExecutor to put drained and exception core in cpp file
KINFRA-1128 In EndpointJobInfoWrapper mark multi-head jobs as done so we don't get an error in dtor
KINFRA-1141 Removed unnecessary try/catch and re-throw in IDb add_objects that was swallowing exception - Additional logging, comments and unit test
KINFRA-1141 Retry pinning of IndexedDb objects when concurrently being evicted under heavy ingest - propagate resource acquisition failures back up to the workers instead of silently failing - removed extra file size checks and added more error checking
KINFRA-1145 Change minimum tcs_per_tom to 8 for >= 4 toms_per_rank
KINFRA-1148 Cancel jobs on worker ranks which haven't started execution yet
KINFRA-1157

Clear worker and request message from job map after worker completes

KINFRA-1157 Fix worker/JM deadlock when clearing workers
KINFRA-1160 bin/gts.sh ignore docker pids when on host
KINFRA-1160 gpudb-httpd.sh ignore docker pids when on host
KINFRA-1161 Fix race condition reallocating join stencils
KINFRA-1163 Fix segfault when unable to allocate PKIndex in ctor
KINFRA-1172 Fix broken /alter/user set resource group security check - System admin users can set resource groups - Non-system admin users cannot set resource groups - admin user's resource group cannot be set
KINFRA-1179 Do not allow alter/table change_column action on materialized_view
KINFRA-1181

Fix valgrind error by explicitly clearing (destroying) GaiaSetData's m_chunk_data vector before m_original_set_id is destroyed

KINFRA-1182 Fix worker infinite loop if cancelled prior to WI scheduling
KINFRA-1185 Add show_resource_statistics to request->response mapper
KINFRA-1186 Check if persist copy of vectors is up to date when copying to cold storage
KINFRA-1186 Fix compilation warning from previous commit
KINFRA-1186

Fix compressed vector metadata and update geometry vector, on append to mark cold storage as dirty

KINFRA-1186 Reset geometry vector cold storage m_backlog
KINFRA-1187 Do not sort UDF input column names
KINFRA-1193 Fixed 'unable to reallocate multiple groups pinned' error
KINFRA-1193 Fixed race condition appending to refreshed materialized view
KINFRA-1194

Don't treat filter views as materialized views for security checks

KINFRA-1196 GlobalManager don't repub rank_info received with same version, GlobalRegistry only resend same version every 5 times to correct GM
KINFRA-1196 GlobalRegistry use map to track repeat sends to handle host_info, print member name in send_member_update
KINFRA-1196 GpudbExeProcess for ranks asks GM for latest rank_info_cluster to fix changing number of ranks when Host Manager is running
KINFRA-1196 Init CommunicatorUtils::m_activeRanks to false so we don't clear hung messages when another rank is 'starting_systems' making r0 hang waiting for a response to the show_types broadcast
KINFRA-1211

Fix 'unknown column' during unpivot refresh

KINFRA-1214

Initialized chunk_id from persist

KINFRA-808

Added changes that should have been in last commit in test_utils.py for sorting within assertEQ

KINFRA-808

Make test_alter_table_chance_column.py do sorted column comparisons so tests pass during rebalancing

KINFRA-909

In EndpointJobInfoWrapper fix default timeout for solo job blocker (was 12 seconds, meant to be 120)

KINFRA-909

Make /visualize/isochrone block solo jobs, same as other graph endpoints

KOLAP-1189

ipv4_part function index starts from 1 instead of 0 for consistency, and some other small code optimizations

KOLAP-1124

Fixed sql.max_view_nesting_levels spacing issue--breaking doc gen

KOLAP-1189

Added CHARN_IPV4_CAST functions

KOLAP-1189

Added charN_to_ipv4 cast functions

KOLAP-1189

Added INT_IPV4_CAST and LONG_IPV4_CAST functions

KOLAP-1189

Added IPV4_LONG_CAST functions

KOLAP-1196

Fixed tests work for different sql planner configurations

KOLAP-1221

Fixes for WKT inserts from ODBC (using Multi-head)

KOLAP-1242

Enhanced filter plan for Geo-joins to display the No. of chunks processed and Total No. of chunks

KOLAP-1253

Added gpudb.conf parameter enable_predicate_equi_join (default false) that when set to true enables the predicate-equi-join filter-plan

KOLAP-1253

One predicate equijoin when there are multiple equality predicates between the same two tables

KOLAP-1258

Explain and describe support for logical views

KOLAP-1278

Added tests for logical view row count validation. Expect right answer for alter view when multihead is on

KOLAP-1295

Fix crash in test_create_projection when run under multiple toms due to bad memory access when shrinking a dict vector down to 0

KOLAP-1295

Fix resharding dict column values during /create/projection resharding

KOLAP-1295

Projections retain dict encoding; if a source column is dict encoded the destination column will be too. The option 'preserve_dict_encoding' is now deprecated and ignored

KOLAP-1331

Additional optimization involving SpaceEfficientVector

KOLAP-1331

Attempt to optimize gathering the chunk indices

KOLAP-1331

Further optimizations for SpaceEfficientVector (for JoinedStencil::get_pk_indices_regular)

KOLAP-1331

Reduced checkPinnedRead calls in inner loops if we already know that stencil is pinned

KOLAP-1338

Disallow all modify operations on logical views

KOLAP-1338

Do not recreate logical views at worker ranks during gpudb restart

KOLAP-1339

Add test to recreate table using show table output

KOLAP-1345

Fixed problem with window function when alias used on shard key - make sure all references to columns are translated to aliases in the projection - Simplify the logic to clarify the process of resharding/replicating projections

KOLAP-1346

Check in schema for internal retain_chunks option

KOLAP-1346

Restore query graph options

KOLAP-1349

Fix /show/table endpoint to return all matches instead of exact match

KOLAP-1357

Show resource group and explain enhancements

KOLAP-1361

Fix "compute_request requires a shard key" error - recognize that if the source table is replicated we can compute all window functions on it before resharding to a requested shard key

KOLAP-1380

Relax name validation in /alter/table, enable name validation for collection name in /create/table

KOLAP-1380

Reserve ROOT and KINETICA key words and avoid user to create objects with reserved names

KOLAP-1381

Fix the test so it is actually applying the expression to filter out the exact record we want

KOLAP-1382

Update to improve testing of sharding

KOLAP-1389

Added PrimaryURL support to ODBC Driver

KOLAP-1393

Increase query planner defaults for higher concurrent query needs

KOLAP-1393

Increase query planner defaults in conf.ini

KOLAP-1393

Update min_http_threads to 8 and max_http_threads to 512

KOLAP-1396

Make sure result_distribution is part of explain verbose

KOLAP-1396

Result table distribution information in the sql explain verbose

KOLAP-1405

Modified SpaceEfficientVector::assign to not clear the vector avoiding changing the capacity and mode

KOLAP-1409

Add between function for all comparable data types - test for datetime added to test_between2d_join_filter.py

KOLAP-1409

Added Coding between - nothing coded enabled - Add dummy between(column,lower_bound,upper_bound,lower_bound_le,upper_bound_le) to be detected by FilterPlanner to implement between2d plan - FilterPlanner::extract_between_2d_join_plans detects between function call and create FilterPlan type BETWEEN_2D_HOIN - EquiJoinKernel - begin implementing between2d_join_func_single - Worker_equi_join_filter - preparing to call between2d_join_func_single for BETWEEN_2D_JOIN plans - Worker_join_filter - case BETWEEN_2D_JOIN added to case EQUI_JOIN 

KOLAP-1409

between-2d filter on equi-join disabled until it compiles correctly in cuda build

KOLAP-1409

between-2d filter plan step after equi-join works

KOLAP-1409

between2d filter plan following block stencil reworked to use lexicographical

KOLAP-1409

First working version of between2d filter-plan - lightly tested - tested on bitmask stencils but not equi-join stencils

KOLAP-1409

Fixed EquiJoinKernel.cu:encode function exclusive_scan

KOLAP-1409

Modify BlockStencil to support BetweenVar filter plan

KOLAP-1409

Tested and fixed problem with between2d filter when there are gaps in the rhs blocks

KOLAP-1409

Worker_equi_join_filter calls between2d_join_func_single for BETWEEN_2D_JOIN filter plan type

KOLAP-1415

Fix for EXPLAIN working with INSERT

KOLAP-1417

Added time_int/long_cast functions

KOLAP-1418

Added tiering mechanism to the new member m_lengths1 of BlockStencil. - Modified the BlockStencil Tiered tests considering the new member m_lengths1

KOLAP-1419

Fixed crash with tier predicate transformation

KOLAP-1420

Change the default conf setting for force_host_filter_execution from 0 to false

KOLAP-1421

Avoid accessing type info with invalid attribute name

KOLAP-1424

Fixed bug in bindMinMaxAggregates

KOLAP-1426

Fix explains for grant/revoke statements and explain analyze for create/clear tables

KOLAP-1431

Geometry columns could not be part of an equi-join

KOLAP-1432

Calculate::chunkerFactory now sets var_set_indices to all_indices if all_dimensions are requested - overly conservative but need to pass variables to avoid this * JoinedSetChunker sets var_set_indices to all_indices if no var_indices were specified - again overly conserative Note - var_set_indices is used to determine which chunks are chunk-locked - being conservative with these does not change behavior and has modest impact on concurrency * added lock checker to GaiaSetDataVector and GaiaSetDictVector pinObject member functions * JoinedSetChunkLock(JoinedSet,chunk_index) constructor now handles primary key set indices correctly * Important locks   - Worker_primary_key_join_filer chunk locks all primary key chunks - Worker_aggregate_group_by_internal incremental update step to accumulate into existing pks now chunk locks all chunks - Worker_create_union JoinedSetChunker locks all set_indices chunks - was locking none - Worker_get_records_by_column_intenal locks all set_indices chunks rather than the dim_set_indices chunks * Locks added to avoid errors from pinObjects lock cheker - Worker_create_projection locks chunk0 of target set before apply_chunking called - GaiaSet::apply_chunking locks new chunks - GaiaSetChunkData constructor - Worker_insert_result_tables_utils locks new chunks

KOLAP-1434

Handle single quote escape correctly when updating records

KOLAP-1438

Expand a.* and * column names for type to create new flat column-name list to be added to external type

KOLAP-1439

Added recomputation of bounding box (Min/Max) of WKT columns after database restart

KOLAP-1441

Fixed long to time conversion bug

KOLAP-1443

6.2  to master. Make rollup() intermediate work tables non-persisted

KOLAP-1444

Set logical view row count as 0 since the workers do not send the row counts for the views

KOLAP-1447

Allow empty collection name in /aggregate/unique endpoint

KOLAP-1455

Change WARNs in GPUManager to ERROR_AND_THROWN

KOLAP-1455

Throw error in GaiaSetDataVector::copy_to if offset_bytes+num_bytes > bytes_per_element*size_elements

KOLAP-1456

Hold chunk_vector_read locks while determined if an incremental refresh can be done and then doing it

KOLAP-726

re-write find_datetime_specifier_code function in parser

KUI-710

restore code that returns the partition_definitions_json - seems to have been eliminated by a previous merge

KUI-721

Fix dashboard toolbar overflow for 7.0

KUI-728

Add ability to view EXPLAIN plan additional info in full browser window

KUI-728

Add popup for table info displaying columns and types/properties

KUI-728

Add row count info to nodes when applicable, fix layout alignment, update heatmap

KUI-728

Add spinner to run and analyze explain plan as it could take some time to run

KUI-728

Add support for EXPLAIN plan additional info tooltip for create joins

KUI-728

Add support in KiSQL screen for EXPLAIN feature

KUI-728

Auto-fit generated explain plan graph in window

KUI-728

Enhanced explain plan runtime heatmap

KUI-728

Fix duplicate table nodes

KUI-728

Fix explain verbose issue with DDL statements

KUI-728

Fix row count formatter

KUI-728

Improve error handling for queries that aren't currently supported by EXPLAIN

KUI-728

Improve how user can opt for VERBOSE option on Run & Analyze

KUI-728

Improve layout/formatting of EXPLAIN plan additional information fullscreen view

KUI-728

Make table_name output table for /append/records endpoint operations

KUI-728

Switch PHYSICAL to VERBOSE for default explain plan

KUI-728

Tweak graph layout algorithm for better results

KUI-728

Update EXPLAIN plan graph to handle tables in plans that do not physically exist

KUI-735

Improve display of resource statistics data

KUI-735

Update styling of Resource Statistics page

KUI-750

Convert table distribution chart to d3.v5.js

KUI-759

Update drivers download list (remove Mac)

KUI-761

Change table info modal DDL implementation to use /execute/sql endpoint instead of KiDDL

KUI-778

GAdmin log report generator not working for rolling logs

 

KNOWN ISSUES:

None currently

  


Should you have any questions or concerns, please visit our support page, official documentation page or email us at support@kinetica.com

 
Was this article helpful?
0 out of 0 found this helpful
Have more questions? Submit a request

0 Comments

Article is closed for comments.