Skip to content

Latest commit

 

History

History
379 lines (368 loc) · 41.8 KB

File metadata and controls

379 lines (368 loc) · 41.8 KB

Change log

Generated on 2026-03-10

Release 26.02

Features

#13381 [FEA] Add support for iceberg partition calculation.
#14137 [FEA] Switch from Java 17 release to Java 8 target for Spark 411 shim.
#14083 [FEA][AUDIT][SPARK-52921][SQL] Specify outputPartitioning for UnionExec for same output partitoning as children operators
#14056 [FEA] Add support for Spark 4.1.1
#13388 [FEA] Add support for iceberg void transform.
#13389 [FEA] Add support for iceberg identity transform.
#13382 [FEA] Add support for iceberg bucket transform.
#9080 [FEA] Support sha2
#13935 [FEA] Add support for iceberg view.
#14066 [FEA] Set correct iceberg table property for spark rapids.
#13750 [FEA]Drop Spark 3.2.x support
#13882 [FEA] Support spark 3.5 + iceberg 1.9.2
#13794 [FEA] Update rapids JNI, private and hybrid dependency version to 26.02

Performance

#14131 [FEA] Enabled Kudo GPU reads by default
#13715 [FEA] Expand gpuSpillTime and gpuReadSpillTime metrics to include full spill operations
#13812 [PERF] Unnecessary data transformation steps in GPU aggregations

Bugs Fixed

#14267 [BUG] Spark 3.5.3 delta reads slower then 3.4.1
#14262 [BUG] NDS query66 unmatched results error in Spark 4.1.1 + ANSI run
#14233 [BUG] test_parquet_testing_valid_files for parquet-testing/data/null_list.parquet falls back
#14197 [BUG] LHA test fails with cudaErrorInvalidResourceHandle during CSV reading on L40S cluster
#14215 [BUG] Test Failure: SparkNumberFormatException - [CAST_INVALID_INPUT] The value '+1.2' of the type "STRING" cannot be cast to "INT" because it is malformed.
#14218 [BUG] Build errors on databricks with shuffle v2 phase 2
#14134 Change calls to cuDF partitioning apis to account for the numRows offset
#14174 OOM when a single shuffle partition exceeds 2GB in RapidsShuffleThreadedWriter
#14201 AB-BA deadlock between SpillableHostStore and SpillableHandle during shutdown
#14196 [BUG] Iceberg identity partition fallback tests failing after GPU support implementation
#14099 [BUG] IllegalStateException during Shuffle Write when Join result exceeds 2GB string limit.
#14037 [AUDIT] [SPARK-52962][SQL] BroadcastExchangeExec should not reset metrics
#14188 [BUG] JDK11 Nightly Build: DatasetSuite test failures for RelationalGroupedDataset toString format
#7520 [AUDIT][SPARK-42045][SQL] ANSI SQL mode: Round/Bround should return an error on tiny/small/big integer overflow
#14179 Merger thread can become zombie when task is killed, blocking subsequent tasks
#14096 [BUG] Window function tests fail with GPU/CPU string value mismatches in min aggregations in cuda13
#13954 [BUG] gpu kudo does not make its inputs spillable
#14030 [BUG] Iceberg GPU result is diff from CPU when doing merge
#13365 [BUG] NDS query57 completed with "IllegalArgumentException : Invalid kudo offset buffer content" error on dataproc
#14092 [BUG] Fix incorrect Iceberg-write row in supportedDataSource.csv‎
#14075 [BUG] Iceberg integration tests fail with pyspark.errors.exceptions.captured.UnsupportedOperationException: Creating a view is not supported by catalog: spark_catalog
#14078 [BUG] SizeInBytes metric display incorrect for data sizes exceeding TB
#14025 [BUG] Integration tests timeout on Dataproc 2.1-debian11: Job 0 exceeded 3600 second limit
#13629 [BUG] error using UCX shuffle with Spark 4.0
#14057 [BUG] Spark 4.0.0 master/worker startup failure: ClassNotFoundException and spark-class script error
#14043 [BUG] test_avg_divide_by_zero failed for OSS Spark 3.3.0 and across all Databricks runtime versions
#13739 [AutoSparkUT] Read row group containing both dictionary and plain encoded pages - Missing GPU verification test
#13760 [AutoSparkUT]DATE_FROM_UNIX_DATE test case in RapidsDateExpressionsSuite threw java.lang.ArithmeticException: integer overflow
#13907 [BUG] join on the condition cast <row> to <boolean> brings errors on the GPU engines
#13953 [BUG] Test Failure: SPARK-24788 RelationalGroupedDataset.toString missing 'type: GroupBy' in output spark330 JDK11
#13914 [BUG] An empty table crossJoin another table brings errors on the GPU engines
#13765 [AutoSparkUT] SPARK-38237 shuffle distribution test needs GPU-specific implementation
#13892 [BUG] CUDF UDF tests fail with kvikio symbol undefined error: _ZN6kvikio12RemoteHandle5preadEPvmmmPN2BS11thread_poolE 26.02 cudf-nightly
#13899 [BUG] Race condition in host-to-disk spill due to premature disk handle exposure
#13820 [BUG] NullPointerException - aggregate.TypedImperativeAggregate.merge when final agg on CPU

PRs

#14385 hotfix release 2602.1 [skip ci]
#14336 Update changelog for the v26.02 release [skip ci]
#14328 Writing field id when writing iceberg's data file
#14255 Update changelog for the v26.02 release [skip ci]
#14254 Update dependency version JNI, private, hybrid to 26.02.0
#14271 Fix combining small files when reading Delta tables using multi-threaded reader
#14241 [DOC] update for download page 2602 release [skip ci]
#14264 Fix GpuHashAggregateExec outputPartitioning for aliased grouping keys
#14243 [SPARK-54220] Xfail null_list.parquet for Spark 4.1.0+ due to array inference
#14226 Disable RAPIDS Shuffle Manager when spark.shuffle.checksum.enabled is true
#14230 Fallback to CPU for hash joins with struct keys having different field names
#14206 Spark-4.1.1: Resolve integration tests in map_test.py and bloom_filter tests
#14164 Align GpuUnionExec with Spark 4.1's partitioner-aware union behavior
#14120 [FEA] Add support for Spark 4.1.1
#14217 Add tests for dml operations after schema evolution.
#14175 Fix OOM when shuffle partition exceeds 2GB in threaded writer
#14202 Fix AB-BA deadlock between SpillableHostStore and SpillableHandle during shutdown
#14204 Fix iceberg identity test failure.
#14189 BroadcastExchangeExec should not reset metrics
#14192 Fix JDK version diff in RelationalGroupedDataset
#14183 Support for Iceberg identity partitioning
#14182 Artifactory credentials for wget used in spark-premerge-build.sh
#14180 Fix merger thread deadlock when task is killed
#14139 Add withRetry to GpuBatchedBoundedWindowIterator
#14166 Fix failed cases due to Spark 41x changed the default mode from unsafe to safe
#14170 Update authorized users
#14161 Add in the proper output ordering and partitioning to GpuWindowLimitExec
#13724 rapids shuffle manager V2 phase 1: writer use as much memory as allowed and pipelined write
#14157 [AutoSparkUT]Add Dataset, DataFrameFunctions and ColumnExpression suites
#14151 [AutoSparkUT]Enable several Spark UT suites
#14141 [AutoSparkUT]Enable several UT Suites
#14125 enable GPU kudo reads by default
#14003 Add in basic GPU/CPU bridge operation
#14130 [AutoSparkUT]Add RapidsCollectionExpressionsSuite
#13995 Add a debug option to check if memory allocation is covered by retry framework
#14031 Fix Iceberg data corruption when meets null in the condition in merge process
#14095 Fix Scala 2.13 compilation warnings and enforce stricter checks
#14035 Allow for AST join build side selection and add some heuristics
#14089 Add length check for materializing to host memory buffer from DiskHandle.
#14001 Supports all types for Iceberg bucket transform
#14093 Fix incorrect Iceberg-write row in supportedDataSource.csv
#14101 Migrate pre-merge CI image usage to new artifactory instance
#14100 Add Spark version notice to generated documentation
#14091 [SparkUT]Fix a from-json case to expect a different exception
#14094 Fix compile error in HashFunctions.scala
#14038 [FEA] SHA-2 hash support.
#14087 Delete useless UT suites which run on CPU only
#14086 Remove the 3rd DataType parameter from all the calls to GpuColumnVector.from(GpuScalar, int, DataType)
#14082 Fix Iceberg view test failures in CICD
#14070 Kudo supports schema check when serializing batches
#14080 Set JDK 17 as the default for nightly builds across both scala2.12 and scala2.13
#14079 Fix metric display for data sizes exceeding TB (#14078)
#14010 support withRetry with split for GPU shuffle coalesce
#14071 [DOC] update supported spark versions [skip ci]
#14033 [BUG] Fix initialization order NPE for RapidsShuffleManager in UCX mode for Spark 4+
#14042 Add Iceberg view test
#14058 Update documentation to reflect the shims updates [skip ci]
#14039 Add support for docs generation in buildall
#13975 support withRetry with split for shuffle exchange exec base
#14044 Adds ignore_order for groupBy agg test that returns multiple rows
#14026 Add in the missing RmmSpark calls for the coalescing reader
#14029 Increase executor memory for iceberg tests to avoid OOM error [skip ci]
#14022 Add RapidsDateFunctionsSuite
#13993 Update plugin scala212 to build 330+ only
#14024 Update actions/setup-java@v5 [skip ci]
#13986 Add iceberg 1.9.2 support.
#14008 Fix auto merge conflict 14007 [skip ci]
#13982 Add parquet mixed-encodings test
#13994 [SparkUT]Add try-catch on dataframe.collect in UT framework
#13997 Add pmattione to blossom: Attempt 2 [skip ci]
#13992 Building scala213 plugin for spark350+
#13987 Add layer of indirection when converting expressions to the GPU
#13981 Add in support for getting SQL metrics from expressions
#13983 Support custom parallelism in non-default test modes [skip ci]
#13964 Fix a join where only a single column is used as a condition
#13970 Make the GPU UDF has different name than the CPU one
#13933 [DOC] fix dead link in testing page [skip ci]
#13845 Add a non strict mode for lore dump
#13959 [SparkUT]Check Java version to decide the expected string in one case of DataFrameAggregateSuite
#13955 [DOC] Update RapidsUDF output types with decimal 128 [skip ci]
#13938 Fix a special case in limit where it could return an empty batch with the wrong number of columns
#13945 Set WONT_FIX_ISSUE cases in RadpisDataFrameWindowFunctionsSuite
#13947 Fix auto merge conflict 13946 [skip ci]
#13936 Add testRapids case to match GPU execution in RapidsDataFrameWindowFunctionsSuite
#13905 Refine GpuTaskMetrics over SpillFrameWork
#13931 Use strict priority in conda process [skip ci]
#13911 [AutoSparkUT]Enable RapidsCsvExpressionsSuite & RapidsCSVInferSchemaSuite
#13900 fix race condition due to premature disk handle exposure
#13902 [AutoSparkUT]Add RapidsCsvSuite
#13875 Persist the bufConverter of TypedImperativeAggregate into LogicalPlan instead of PhysicalPlan
#13883 [AutoSparkUT]Enable several Spark UT suites
#13887 Fix auto merge conflict 13877 [skip ci]
#13857 Use shim to identify whether DataWriting is supported for LoRe
#13819 Add the missing spark 357 version to GpuWriteFilesUnsupportedVersions
#13777 [AutoSparkUT] Migrate DataFrameNaFunctionsSuite tests to RAPIDS
#13796 Update dependency version JNI, private, hybrid to 26.02.0-SNAPSHOT [skip ci]
#13767 [AutoSparkUT] Migrate DataFrameComplexTypeSuite tests to RAPIDS
#13762 [AutoSparkUT]Enable Spark UT DateExpressionsSuite
#13795 Bump up version to 26.02 [skip ci]
#13790 Use wildcard mark to import all Rapids test suites migrated from Apache Spark Suites

Release 25.12

Features

#13520 [FEA] DML Support for iceberg
#12229 [FEA] Support insert statement for iceberg table.
#13419 [FEA] Enable iceberg insert support by default.
#13383 [FEA] Add support for iceberg truncate transform.
#12719 [FEA] Run iceberg nds power run tests.
#13548 [FEA] Accelerate deletes on clustered tables
#13547 [FEA] Accelerate updating clustered tables
#13386 [FEA] Add support for iceberg day transform.
#13387 [FEA] Add support for iceberg hour transform.
#13384 [FEA] Add support for iceberg year transform.
#13385 [FEA] Add support for iceberg month transform.
#13546 [FEA] Accelerate merging clustered tables
#9079 [FEA] Support map_from_entries
#13636 [FEA] Support merge command for iceberg's merge on read mode.
#13635 [FEA] Support update command for iceberg's merge on read mode.
#13634 [FEA] Support delete command for iceberg's merge on read mode.
#13705 [FEA] Support GBK encoded data
#13470 [FEA] Support Spark 3.5.7
#13468 [FEA] Update rapids JNI, private and hybrid dependency version to 25.12.0-SNAPSHOT
#13339 [FEA] Support Delta Lake 4.0.0
#13525 [FEA] Support merge statement for iceberg's copy on write mode.
#13524 [FEA] Support Update statement for iceberg's copy on write mode.
#13523 [FEA] Support Delete statement for iceberg's copy on write mode.
#13522 [FEA] Support RTAS for iceberg.
#13605 [FEA] Support insert into overwrite static for iceberg.
#13604 [FEA] Support insert into dynamic overwrite for iceberg.
#13521 [FEA] Support CTAS in for iceberg.
#13110 [FEA] Support DeltaDynamicPartitionOverwriteCommand for deltalake 3.3.x

Performance

#13266 [FEA] Upgrade to UCX 1.19
#13462 [TASK] Add microbenchmarks for Delta
#12968 [FEA] Use kudo kernels in the plugin
#12919 [FEA] Triple buffering: traffic control by a customized thread pool
#13568 [TASK] Analyze Deletion Vector read performance against CPU FileSourceScanExec with Predicate Pushdowns

Bugs Fixed

#14009 [BUG] Delta Lake deletion vector tests failing on Spark != 3.5 due to version incompatibilities
#14004 [BUG] test_delta_deletion_vector_read_drop_row_group fails with #13980
#13950 [BUG] Delta parquet files are not split when they can be
#13873 [BUG] join an empty table brings errors on the GPU engines
#13917 [BUG] NDS2 Delta Microbenchmark Consistent Test Failures on SPARK2A Cluster
#13881 [BUG] Iceberg REST catalog integration tests timeout after 5 hours
#13956 [BUG] iceberg AWS S3 Tables test timeout 24 hours
#13940 [BUG] [Iceberg] After update MoR run select collect raise column not match
#13854 [BUG] run_other_join_modes_tests Integration tests: 1121 broadcast join failures and timeout the CI
#13885 [BUG] New added iceberg cases causes nightly integration tests timeout after 12 hours
#13788 [BUG] Scala test 'AsyncCpuTask task priority' failed UT
#13798 [BUG] GpuLoreSuite test failure: GpuInsertIntoHiveTable LoRE replay produces no data (Spark 3.5.7)
#13848 [BUG] test_delta_merge_sql_liquid_clustering fails with Delta 4
#13860 [Test] ERROR: Could not find a version that satisfies the requirement pyspark-client==3.5.6 and 3.5.7
#13852 [BUG] Iceberg merge tests failing after Delta clustered table update support
#13855 [BUG] Scala test GpuInsertIntoHiveTable with LoRE dump and replay failed in CI
#13718 [BUG] Spark Connect smoke test uses a pip package pyspark[connect] with jars
#13731 [BUG] .select() query with left_outer join() brings errors when using the GPU engines
#13824 [BUG] ArrayIndexOutOfBounds when reading iceberg table.
#13826 [BUG] markdown link check fails 404 of logging_resource_adaptor
#13660 [BUG] dataproc serverless failed: AssertionError: Assertion should have executed on driver
#13737 [AutoSparkUT] conv - Base conversion function fails on GPU
#13738 [AutoSparkUT] log10 - Function has different output on GPU
#13740 [AutoSparkUT] log2 - Function has different output on GPU
#13741 [AutoSparkUT] binary log - GPU returns NaN instead of null for negative base
#13689 [BUG] Failed to authenticate in iceberg rest catalog tests
#13800 [BUG] test_csv_read_gbk_encoded_data fail Part of the plan is not columnar class of DB 14.3 runtime
#13779 [BUG] Divide-by-0 exception in Delta optimized write
#13771 [BUG] RapidsMathExpressionsSuite fails on Spark 330: atanh FAILED
#13774 [BUG] pool tasks can mistakenly be registered as dedicated
#13732 [BUG] Small per task leak in RmmSpark task metrics
#13716 [BUG] A special select query with the where expression brings inconsistency when using the CPU and GPU engines
#13708 [BUG] A special select query with the join on expression brings inconsistency when using the CPU and GPU engines
#13690 [BUG] java.io.NotSerializableException: com.nvidia.spark.rapids.RapidsConf occurred in iceberg rest catalog tests
#13664 [BUG] some hash_aggregate integration tests failed with kudo.serializer.mode=GPU
#13651 [BUG] plugin generate different output from CPU run
#13694 [BUG] databricks shims failed compile: not found: value StatsExprShim
#13662 [BUG] GpuRetryOOM in KudoGpuSerializer splitAndSerializeToDevice
#13661 [BUG] Multiple NDS queries failed due to unmatched results on Dataproc
#13612 [BUG] Kudo CPU serializer overflows with columns in a customer join query
#13641 [BUG] test_delta_overwrite_dynamic_missing_clauses fails with "RapidsDeltaWrite is not found in any captured plan"
#13648 [BUG] delta_lake_merge_test:test_delta_merge_match_delete_only failed mismatch cpu and gpu outputs
#13652 [BUG] mortgate_test and dpp_test failure in CI
#12246 [BUG] The plugin can use a different GPU to validate the GPU architecture when multiple GPUs are equipped
#13659 [BUG] validateGpuArchitecture sometimes looks at the wrong GPU
#13647 [BUG] AssertUtils is not a member of package com.nvidia.spark.rapids
#13623 [BUG] Iceberg s3table cases failed multiple different errors
#13059 [BUG] Delta Lake writes are not being checked for CPU fallback in integration tests
#13607 [BUG] DB builds are failing in premerge.
#13583 [BUG] shuffle dockerfile build failed: UCX has no cuda13 release for rocky/centos
#12891 [BUG] NullPointerException at ai.rapids.cudf.HostColumnVectorCore.getByte

PRs

#14077 [DOC] Update document issue [skip ci]
#14028 [DOC] update cuda13 related jars in download doc [skip ci]
#14045 Update version to 25.12.1-SNAPSHOT
#13968 Update changelog for the v25.12 release [skip ci]
#14011 [DOC] update for download page 2512 release [skip ci]
#14013 Fixes various issues in 25.12 release branch
#13988 Update dependency version JNI, private, hybrid to 25.12.0
#13980 Delta table scan should be optimized when deletion vectors don't exist
#14002 Skip test_delta_filter_out_metadata_col on
#13991 Drop _tmp_metadata_row_index column from the output of Delta Scan on GPU
#13903 Fix join bug on csv datasources
#13971 Fix a special case in limit where it could return an empty batch with the wrong number of columns (#13938)
#13962 Cut iceberg test cases for remote catalogs.
#13941 Fix Iceberg column not match error when select count after cor update
#13942 Add null safety guards to the conversion methods in FromIcebergShaded
#13926 Fix columnar mismatch bug in iceberg dml when aqe enabled.
#13929 Fix the ordering of join gather maps, and make it harder to break
#13922 Remove low priority iceberg it test cases to save test time [skip ci]
#13919 fix race condition due to premature disk handle exposure
#13891 Enable iceberg write by default, and disable dml iceberg operation on mor table by default.
#13910 Fix the unstable unit test case: ResourceBoundedExecutorSuite
#13797 Supports Iceberg truncate transform
#13886 [WAR] Remove ICEBERG_ONLY tests from default mode [skip ci]
#13872 [Backport to 25.12] add missing spark 357 to GpuWriteFilesUnsupportedVersions
#13849 Fix test_delta_merge_sql_liquid_clustering for Delta 4
#13867 Use pyspark-client only for Spark 4.x [skip ci]
#13864 [WAR] Move run_other_join_modes_tests out of DEFAULT mode [skip ci]
#13859 Delete support for clustered tables
#13856 Fix test cases: Iceberg datetime transforms
#13778 Use pyspark-client for the Spark Connect smoke test
#13822 Support update for Delta clustered tables
#13726 Supports year/month/day/hour partition transforms for Iceberg
#13830 Fix ArrayIndexOutOfBounds in iceberg reader.
#13719 Upgrade ucx to 1.19.1-rc2
#13814 Support merging clustered tables for Delta IO
#13717 Add different join strategies, join logging, and heuristic confs
#13806 Support GpuMapFromEntries to prevent fallback
#13755 Add update/merge command support for iceberg's merge on read mode.
#13827 Fix broken link to rmm logging_resource_adaptor source code [skip ci]
#13818 Fixes an issue were we were not properly checking log params and nullable output
#13803 Fix a GBK test failure on DB 143+
#13725 Add delete support for iceberg's merge on read mode.
#13792 Add doc for RANDOM_SELECT and oom_injection_mode in integration tests readme [skip ci]
#13780 Add missing handling for the 0 input partition count in computing partition count in GpuOptimizeWriteExchangeExec
#13742 Add Spark 3.5.7 support
#13773 Add an option to run a random subset of integration tests
#13786 Return true when comparing 2 variables if both isNaN in Unit Test
#13761 [AutoSparkUT] Migrate DataFramePivotSuite tests to RAPIDS
#13775 Ensure that task threads are the only ones registered as dedicated
#13733 Calls removeTaskMetrics to remove task metrics in RmmSpark after task completion
#13688 Use new API to do Iceberg partition.
#13768 Refactor RapidsTestSettings imports to one suite per line
#13753 [AutoSparkUT] Add RapidsParquetEncodingSuite migration
#13748 [AutoSparkUT]Enable Spark UT RapidsMathExpressionsSuite, RapidsMiscFunctionsSuite on 330
#13735 Align with CPU behavior for the null handling in GpuInSet
#13656 Add op time metric to Hybrid Scan
#13721 Enable Delta Lake 4.0.x integration tests in nightly CI/CD jobs
#13710 Enable Delta Lake 4.0.x support for Spark 4.0.1
#13489 Support kudo GPU shuffle reads in the plugin
#13711 define usesKudoGPUSlicing override got GpuSinglePartitioning
#13691 Support the null-aware anti join in GPU broadcast hash join
#13413 Fix core dump in MemoryCleaner
#13707 Refactor DeltaDynamicPartitionOverwriteCommand and enable Delta 4.0 integration tests
#13701 Add StatsExprShim in databricks shim
#13700 Support ChangeLog for the new branch model [skip ci]
#13695 [DOC] update the download doc [skip ci]
#13692 Add missing javadocs in delta-33x-40x refactored code [skip ci]
#13666 Add common infrastructure and Spark 4.0 compatibility for Delta Lake 4.0.0 support
#13663 Add withRetry for gpuSplitAndSerialize in GPU kudo
#13643 Add allow_non_gpu_delta_write_if to conditionally allow CPU fallback for delta writes
#13130 Resource-bounded MultiFileCloudPartitionReader
#13657 Add support for update iceberg table with copy on write mode.
#13670 Revert "Temporarily disable test_dpp_reuse_broadcast_exchange and mor…
#13639 Support delete command for iceberg's copy on write mode
#13658 move validateGpuArch check to after initializeGpuAndMemory
#13644 Add FileFormat, Catalog, Provider, and RowIndexFilters base classes and refactor delta-33x
#13654 Move AssertUtils to the api submodule
#13655 Temporarily disable test_dpp_reuse_broadcast_exchange and mortgage_test given cuDF issue
#13650 Mitigate mamba hang on Rocky linux [skip ci]
#13642 Add Create/Optimize table command base classes and refactor delta-33x
#13584 Use assertInTests for cases with potential side effects, or expensive calls
#13637 Update automerge pattern to release/* [skip ci]
#13619 Add replace table as select for iceberg.
#13614 Migrate most remaining NvtxRanges to use NvtxId/NvtxRangeWithDoc
#13626 Add Update/Merge command base classes and refactor delta-33x
#13620 Add shims layer, Delete command base classes and refactor delta-33x Delete
#13586 Assert RapidsDeltaWrite in integration tests
#13613 Add Delta Lake 4.0.x module skeleton and build infrastructure
#13611 Add insert static overwrite support for iceberg.
#13606 Add insert overwrite dynamic support for iceberg.
#13595 Add create table as select support for iceberg.
#13573 Accelerate DeltaDynamicPartitionOverwriteCommand
#13608 fix db GpuBroadcastHashJoinExec CollectTimeIterator args
#13594 Use Cuda 11 ucx build for rocky dockerfiles
#12900 Migrate CollectTimeIterator to use new NvtxIdWithMetrics class
#13599 Fix auto merge conflict 13598 [skip ci]
#13501 Use RapidsFileIO for writing data.
#13572 Upgrade UCX to 1.19-rc1
#13544 GpuBubbleTime: A new metric recording GPU underutilization wall time
#13500 Add Spark Connect smoke test in nightly integration tests
#13399 Add two metrics for the sized join
#13454 log if maxMem is called twice in a task
#13475 Update dependency version JNI, private, hybrid to 25.12.0-SNAPSHOT

Older Releases

Changelog of older releases can be found at docs/archives