Follow

Kinetica 7.0.10.0 Release Notes

Build version: 7.0.10.0

  • 132 Resolved Issues

FIXES: 

KDEVOPS-744 Added yum update and yum cleanup to make-gaia-docker.sh
KDEVOPS-759 Change '--session-command' to '--command' on su in postinst for ubuntu
KDEVOPS-759 Fix custom slice search for Ubuntu
KDEVOPS-759 Fix post install script type-o gpudb-python=>gpudb_python
KDEVOPS-759 Rebuild custom reveal slices in post install
KGEO-677 Added a new solve option to match/graph for only the case where source and destination pairs are of WKTPOINT type, i.e., 'match_batch_solves'
KGEO-685 Change some logs from INFO to DEBUG
KGEO-692 Add comments for VTS and style line rendering shader
KGEO-694 Accurate ShapeProcessor accounting
KGEO-694 Eliminate ShapeProcessor LOD memory limit conf parameter
KGEO-694 Fix VTShapes compile error under gcc6
KGEO-694 Loop on LOD eviction to make sure to exhaust each ShapeProcessor's contents before moving to the next Don't attempt LOD eviction during queries, only attempt from ShapeProcessor threads, since it could take a while Fix potential memory leak of geometry objects if gsgv::appendGeometry fails reallocation
KGEO-694 ShapeProcessor clear() -> clearMem(), use swap to clear std containers, make ShapeAtttributeBase virtual to fix clear()
KGEO-699 Correct initialization of LRU partition list for shape processor memory usage control
KGEO-700 Changed user specific graph errors to correctly return UERR
KGEO-701 Grid functions (st_hexgrid/st_pointgrid/st_trianglegrid/st_squaregrid) can output up to 20 million cells (up from 1 million). Plus much faster grid generation since rank-0 doesn't need to convert the geometry to WKT strings
KGEO-704 Added support to cancel mapmatching jobs
KGEO-706 Added a missing solver option 'allpaths' (documented but not listed)
KGEO-706 Added a new create/graph option 'modify' to modify an existing graph for new inserts and/or deletes
KGEO-708 The graph client will now force reset the graph server after a rank0 kill
KGEO-713 Allow endcap style 'butt' as an alias for 'flat' in st_buffer
KINFRA-1080 Minor fixes to EglTest.cpp
KINFRA-1095 all_test_nose.py now calls show_resource_statistics() not show_table(get_sizes=true) to test if db is up, way to slow if many tables
KINFRA-1095 Fix byte packing of MessageHeader to silence valgrind errors sending the seriaized version of them
KINFRA-1160 Relax gpudb_sysinfo.sh filter docker pids to look for only /docker, not :/docker in cgroup
KINFRA-1598 Fix /show/system/properties reporting of 'request_timeout' even though internally we store it as seconds externally (via config and docs) it is provided in minutes
KINFRA-1605 Add null checks in GaiaSet dtor
KINFRA-1606 Add file 'delete_this_file' in persist for deleted tables, and clear empty or deleted tables directories in GaiaSet::sync_chunks
KINFRA-1608 Fix managed containers to not auto-convert to/from std containers
KINFRA-1609 Check for null input when encoding JSON
KINFRA-1609 CMakeGaiaCuda.cmake -DBOOST_NO_CXX11_VARIADIC_TEMPLATES if cuda8
KINFRA-1609 Make managed per field instead of per type
KINFRA-1609 Rem out last change adding for cuda8 -DBOOST_NO_CXX11_VARIADIC_TEMPLATES to fix linker errors. Revert to fix gcc >5 boost::vector::emplace_back() errors, we make sure that is not #defined in any case in core libs now
KINFRA-1609 Replace Avro encoder and generator
KINFRA-1609 Use MemoryStream instead of StringStream for JSON decoding (to use size instead of null terminator)
KINFRA-1618 Add check for memory reallocation in geometry vectors, before actual insert, in GaiaSet::increase_capacity_if_needed_and_create_add_desc
KINFRA-1618 Correct number of elements for geo vectors pre-allocation evaluation
KINFRA-1618 Fix addl_bytes calc in bulk_insert geom check
KINFRA-1618 Fix GaiaSet compilation in gcc6+
KINFRA-1620 Convert aggregate_group_by_internal_response to managed
KINFRA-1620

Fix cuda8 build error with ManagedVectors in groupby kernel

KINFRA-1620 Fix groupby non-grouped geo aggregates broken during the managed vector conversion
KINFRA-1620 Undo CUDA 8 fix (should no longer be necessary with boost fix)
KINFRA-1628 Fix JobManager/TableLockManager deadlock issue; also fix issue where an unlock could be missed for a stage 1 exclusive lock
KINFRA-1629 JsonWrapper fix errors in loadFromFile unset changed flag after reading from file
KINFRA-1629 JsonWrapper: save json to a temp file, and only when successful, rename to orig file name. Call SaveToFile on all json files directly, instead of dtor, so can throw on error
KINFRA-1636 In JobManager reset shared pointer to messages on worker completion, rather than swapping out message data byte vector
KINFRA-1652 Add test in SetStencilTest
KINFRA-1655 RamPool improvements and cleanup
KINFRA-1664 Use full set data ID for GaiaSetData-related lock and pool names
KINFRA-1741 Clear temp key/index vectors in GaiaSetDictVector
KOLAP-1108 Cherry picked shadow cube changes from 7.0.9
KOLAP-1108 Fix merge conflicts and add tests for shadow cubes
KOLAP-1565 A left join consisting of multiple primary key joins to cross join on the right side with and without a final bitmask1d filter now works and produces the same results as a predicate join with the same joining expressions
KOLAP-1565 Avoid compiler warning about uninitialized chunk_size
KOLAP-1565 Problem described in ticket fixed but fix should be moved from Worker_one_dimensional_join_filter to the Worker_primary_key_join_filter
KOLAP-1584 Fix total_number_of_records when offset is used in the sql query Support /alter/system/properties and /show/system/properties through SQL which can be used to control the sql processing behaviour
KOLAP-1600 Add support for * in /show/sql/proc
KOLAP-1600 Add support to /show/sql/proc to return all sql procedures info
KOLAP-1600 Bypass security checks only for endpoints which allow it
KOLAP-1600 Do not unauthorized users to view sql procedures
KOLAP-1600 Fix show procedure output and add tests
KOLAP-1600 Return multiple procedure names in response
KOLAP-1685 Move locking joined_type_id_to_type_map_lock before creation of out_jsd.  Not sure what this is locking as there are no shared variables in the FilterPlanner but fixes the problem
KOLAP-1695 Support wild-card character '*' in-place of schema/tablename for grant/revoke statements
KOLAP-1712 Set source table ttl to -1 to make sure the append/records input does not expire
KOLAP-1721 Enhance show table output to print table properties using new syntax
KOLAP-1722 Do not drop Materialized view when underlying table is altered with ADD COLUMN
KOLAP-1726 Add a test for persisted projections involving WKTs that was failing with the previous implementation attempt
KOLAP-1726 Allowing the non-charN columns to be persisted during projection. -The non-charN column values are stored in a Managed Vector for accoounting purposes in GaiaSet. -The IndexDB is updated by reading the column values from Managed Vector
KOLAP-1736 Between and Unary filter plans implemented with TimedIndexedWorkItemScheduler
KOLAP-1736 Fixed TimedIndexedWorkItemExecutor to correctly handle thown errors from WorkItem
KOLAP-1736 TimedIndexedWorkItemExecutor will not schedule more workers than there are chunks
KOLAP-1744 Fix the in operation bug by adding CharNInSimplification
KOLAP-1747 Add few join table options to consuming endpoints
KOLAP-1747 Add tests to see the join explain text is shown for all the consuming endpoints
KOLAP-1771 to_char function supports number as string format specifier
KOLAP-1772 Add accelerated parallel-chunk-skip-index-analysis path that processes just the chunks returned by the index
KOLAP-1772 Chunk skip index file is deleted when the table is cleared
KOLAP-1772 Chunk skip index supports date/time/datetime/timestamp datatypes
KOLAP-1772 Clean up chunk skip index implementation
KOLAP-1772 Initial implementation of a 'chunk skip' index. -Index is created via /alter/table with the option 'index_type' set to 'chunk_skip' (the option set to 'column' is the default and that is for our standard attribute indexes. -The index will be populated via column metadata when possible (i.e. which the column min/max values are identical so in that case the column data itself won't need to be read. -The index maintains a map at the GaiaSetData level of unique values to a list of chunk ids that contain that value (i.e. the value 5 is in chunks 0, 3 and 7). -At query time the filter planner will use the chunk skip index to satisfy queries of the form 'column = <value>'. -The index is only helpful when there are tons of chunks and when the speed of our standard chunk skipping becomes costly
KOLAP-1773 Sort get/records/by/column results
KOLAP-1778 Do not cache aggregate chunk results when the request refers to ki_partition()
KOLAP-1780 Fix SQL show procedure statement output to include execute_interval and starting timestamp
KOLAP-1781 Add tracers to keep track of resident set state. Tests/benchmark_shadow_cube.py
KOLAP-1781 Fix vector overrun issue prior to saving predicate join stencil to cache
KOLAP-1782 TIMESTAMPADD returns datetime when input is a valid datetime string format
KOLAP-1783 Check bound values when pass datetime literal string in timestampadd functions
KOLAP-1786 Bump set_index from 32 to 40 bits in gaia_id to avoid issues with sorting once the set_index gets large
KOLAP-1798 Treat and/or expressions with nullable inputs like other operators if both arguments can be shown to have equal null values
KOLAP-1801 Enclose escape character for like function with single-quotes
KOLAP-1801 LIKE operator accepts a third argument, with default escape char of '\'
KOLAP-1811/1812 Disallow refreshing a joined table that is not a part of materialized view. 2. set no_incremental_refresh to true for all joined sets that are not part of materialized views
KOLAP-1813 CONSTANT_TIMESTAMPADDSimplification only needs to handle datetime string, not date string
KOLAP-1814 Fix valgrind error when doing between 2d filter by avoiding the calculation of block areas when the bound_type is not EQUAL_BOUND block areas are only needed for equi-join stencils
KOLAP-1815 Aggregate shadow cube uses ram pool's memory. Fix a circular reference issue in invalidating the aggregate table caches
KOLAP-1816 Test and fix disabling of overlap equi-joins fix flip calculation in FilterPlanner.cpp:accumulate_compound_equijoin_parameters
KOLAP-1817 Add new gpudb.conf parameter 'enable_overlapped_equi_join' to, when set to false, allow a user to force using the predicate-equi-join rather than overlapped equi-joins. Makes it possible when set for the shadow cube to cache BlockStencils
KOLAP-1817 Begin extracting compound-predicate-equi-join plan
KOLAP-1817

begin implementing EquiJoinKernel.cpp:create_multi_equijoin_

block_stencil_for_indices_and_values

KOLAP-1817 Enable COMPOUND__PREDICATE_EQUI_JOIN path in Worker_join_filter and Worker_predicate_join. Add new function to EquiJoinKernel get_multiple_expression_indices_and_values. Fix GPUManager bug that get_null_stencils does not get null_stencils for expressions of attributes
KOLAP-1817 Enable compound-predicate-equi-join and modify tests to avoid comparison until fixed
KOLAP-1817 Fix testing error in test_join_predicate_equijoin.py
KOLAP-1817 Initial implementation of compound-equi-join EquiJoinKernel does the multiple steps using the original algorithms will change to one step mult-attribute algorithm in next pass
KOLAP-1817 Initial simplifications to FilterPlanner and EquiJoinKernel
KOLAP-1817 Move modification of predicate-join filter plans to handle outer join out of loop
KOLAP-1817 Progress toward extracting compound-predicate-equi-join plans
KOLAP-1817 Working version of compound-predicate-equijoin only remove null_stenciled values from first join use block stencil indices to permute values for remaining joins
KOLAP-1824 Add SQL support for chunk skip index
KOLAP-1825 Wrap synchronous worker execution into a utility function which throws exception if the worker fails
KOLAP-1827 GaiaSetBase::hasNonPartitionedFactSet
KOLAP-1827 JoinedSets and aggregate_group_by result sets are now not incrementally refreshed if input fact set is partitioned
KOLAP-1829 Add simple Utils/ConfigParser.h/cpp to better handle ini value reading, gpudb_conf stats_server_ip_address is now defaulted to rank0.host
KOLAP-1829 Add StringUTils::to/fromStringBytesSi/Time() for parsing strings of '10 MB'
KOLAP-1829 Fix StringUtils::fromStringBytesSi() rounding when negative
KOLAP-1831 Fix bug that made non-partitioned GaiaSets not incrementally refreshable
KOLAP-1834 Fix showing shadow filter/aggregate sizes
KOLAP-1835 Make a combo cache entry for pk lookup table and bitmask stencil
KOLAP-1873 Fixed bad check of error condition
KOLAP-1877 Change simplified output to use double-quotes instead of single-quotes; add a basic test
KOLAP-1877 get_records_by_column option 'get_simplified_column_names'
KOLAP-1879 Change test_join_equijoin_ai_on_empty_chunks.py to not clear all tables
KOLAP-1881 Add test_join_kolap1881.py to all_test.sh and join_test.sh
KUI-1053 Add scale bar to Reveal map slice
KUI-1066 Add support for custom WMS param via append query string for additional Map slice layers
KUI-1072 Fix string formatting when column is string nullable
KUI-1072 Scatterplot incorrectly detects numeric columns if they are nullable
KUI-1082 Add legend for Map slice layers and display corresponding fill/line colors
KUI-1086 Map Slice Location Search
KUI-1088 Improve upgradable of custom Reveal slices
KUI-1099 Fix OR alignment on Map slice multi-layer config panel


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.