4.0.13

4.0.13

Release Date: July 16, 2026

Behavior Changes

• The escape handling of LIKE predicates with constant operands (folded on the FE) now matches MySQL 8: SELECT 'a\\b' LIKE 'a\\\\b' returns 1 and SELECT 'a\\b' LIKE 'a\\b' returns 0. Queries that relied on the previous non-MySQL escaping semantics will return different results. #74814
• SHOW [FULL] FUNCTIONS now always includes the isolation property (shared or isolated) in the Properties column of UDFs, so users can tell whether the property is set without recreating the function. #75255
• Iceberg REST catalogs with vended credentials use the table metadata cache again, reverting the earlier cache bypass that sent every getTable() to the REST catalog and caused AWS Lake Formation Rate exceeded failures. Cached tables now renew their credentials on every refresh cycle, and the table cache expiry for REST catalogs is additionally capped at 3000 seconds. #75431

Improvements

• Added checksum protection for shared-data tablet metadata and transaction logs. #74924
• Supported combined transaction log / file bundling for FRONTEND_STREAMING loads. #74460
• Scoped shared-data schema-change job locks to the table to reduce lock contention. #75087
• Batch tablet force-delete marking now acquires the TabletInvertedIndex write lock once per batch instead of once per tablet. #75616
• Added an FE metric for the maximum pending-publish time of committed transactions. #75025
• Added a memory limit check for column upgrades in window operator processing. #75821
• Removed unnecessary per-row seeks in the offsets-only read path of array columns. #75861
• Foreground row-count estimation of Iceberg tables is now computed from manifest metadata without enumerating every data file. #75280
• Addressed security vulnerabilities (CVE): excluded the vulnerable org.jline:jline (jline-remote-telnet) from Hadoop transitive dependencies, and upgraded jackson-databind to 2.21.4. #75066 #75373

Bug Fixes

The following issues have been fixed:

• Incremental scan-range scheduling could recompute a different per-driver layout when reusing a deployed fragment instance, leaving part of the scan ranges unconsumed and losing rows (for example, INSERT from Hive). #74674
• INT96 timestamps nested inside ARRAY/MAP/STRUCT in Parquet files read via FILES() or Broker Load missed the session-timezone conversion and were returned shifted by the timezone offset (top-level INT96 columns were correct). #74868
• The audit log did not record the exported row count of SELECT INTO OUTFILE. #74467
• A strict cast could raise an overflow error from the underlying data of NULL rows, which should be ignored. #74903
• parse_json did not respect the ALLOW_THROW_EXCEPTION setting when handling invalid input. #74976
• First-load statistics collection could not be enabled per table while disabled globally: an explicitly set table property now takes precedence over the global configuration enable_statistic_collect_on_first_load. #74794
• Partial column updates on shared-data tables could crash the BE or silently corrupt data when the tablet schema drifted from the transaction schema. #74005
• Unexpected BE process restarts. #74424
• A BE crash (SIGFPE) in Iceberg truncate/bucket partition transforms when the width or bucket count is zero. #74998
• A BE crash caused by a null driver_executor in FragmentContext::set_final_status. #75030
• A race between transaction begin and autovacuum could delete a still-needed transaction log, permanently wedging the partition's publish ("Both txn_log and corresponding tablet_meta missing"). #74906
• avg(DISTINCT x) was incorrectly rewritten to use a sum/count materialized view, returning wrong results. #75071
• A boundary bug in TopN RANK sorting could produce incorrect results. #75045
• split / split_part / str_to_map with an empty delimiter could read out of bounds on invalid UTF-8 input. #75068
• ALTER TABLE ... MODIFY COLUMN ... AFTER a nonexistent column now returns a clear error message. #75073
• A BE crash (SIGFPE) in mod() / pmod() when computing the type's minimum value modulo -1. #74980
• bar() grew memory without bound on a negative or huge width (potential DoS); such inputs are now rejected with an error. #75143
• The transaction-state callback was not unregistered when a multi-statement stream load task was removed. #75188
• A crash when a query was cancelled during spill partition sorting. #75140
• The query memory limit was not enforced during table function execution. #75179
• Selecting a column named floor or ceil failed at parse time with a ClassCastException. #75241
• A heap-use-after-free in OrderedPartitionExchanger when the previous chunk was mutated downstream. #75279
• Three FE metadata-lock correctness races. #74968
• Load spilling could dereference a missing query context when recording spill metrics. #75236
• ADLS2 ListPaths on storage accounts without hierarchical namespace caused CN crashes and vacuum failures. #75166
• BE/CN JVM metrics emitted invalid Prometheus # TYPE lines. #75240
• JIT code generation truncated LARGEINT literals to 64 bits, producing wrong results. #75137
• A combined ALTER TABLE on an external Iceberg table re-executed already-queued actions. #74036
• A nested-loop join crash caused by a build-side column nullability mismatch. #75343
• Iceberg manifest column statistics are now cached selectively to avoid excessive FE memory usage. #75395
• addPhysicalPartition could create only one physical partition per call, making physical-partition backfill of random-distribution tables extremely slow. #75430
• A SQL injection vulnerability in the information_schema.task_runs predicate lookup. #75520
• A BE crash (SIGSEGV) when multiple UNNEST calls in one query referenced the same array column. #75012
• Incorrect translation of nested dictionary-encoded expressions across exchange nodes. #75246
• The join skew hint was lost when the optimizer duplicated expressions. #68964
• Collecting the tables referenced by a view did not skip CTE references. #74813
• A BE abort on CAST(json AS STRUCT<...>) when a struct field name is not a valid JSON path. #75355
• Temporary Parquet dictionary-code columns could leak to upper layers of the execution plan. #74452
• Incorrect results caused by the sort-column elimination optimization. #74983
• The UNNEST result struct type was not narrowed consistently with its input array element type. #75445
• A race between the FE EOS-cancel and the BE stage-2 deployment could mark a successfully finished query as cancelled. #75009
• Join reorder pruning could drop columns that predicates still referenced. #74791
• The global metadata lock was not released when a metadata image dump failed to acquire some database locks, blocking subsequent metadata operations. #75488
• StringSearch::_pattern was left uninitialized, and the split_debug_symbol log output was incorrect. #75614
• LIKE with the single-character wildcard _ returned wrong results on columns with a GIN inverted index. #75551
• The segment-iterator vector could get positionally misaligned during shared-data primary key index rebuild. #74887
• With file bundling enabled, vacuum could delete a bundle file still referenced by sibling tablets because orphaned bundled segments were not flagged as shared; subsequent publishes then failed with "Object ... does not exist". #75689
• histogram() crashed on a non-positive bucket count; it now returns an error. #75041
• A compaction crash (CHECK failure in JsonMergeIterator) when a flat-JSON column changed from NOT NULL to nullable. #75680
• Common predicate operators were lost when the join tuning guide rebuilt a join. #75773
• An UnsupportedOperationException in ApplyTuningGuideRule caused by an immutable inputs list in OptExpression. #70785
• SHOW PARTITIONS and partitions_meta reported the logical-partition bucket count instead of the per-physical-partition count for shared-data tables. #75734
• A memory allocation exception in NLJoinProbeOperator::pull_chunk crashed the BE instead of failing the query. #75788
• SHOW CREATE ROUTINE LOAD emitted a spurious comma before the first load-property clause, making the output non-executable. #75522
• CTAS (CREATE TABLE AS SELECT) did not accept an ENGINE clause, making CTAS into a Unified catalog impossible. #75771
• OR predicates over null-safe-equal (<=>) join conditions were incorrectly rewritten to UNION ALL, returning wrong results. #75038
• A query cache normalization crash for tables with sub-partitions. #75789
• array_map / transform dropped NULL rows when all non-null input arrays were empty. #75141
• regexp_extract_all fell into an infinite loop on a zero-length capture group. #75798
• Flat-JSON subfield reads returned NULL for keys whose name contains ., which could also cause silent row loss when predicates were pushed down. #75583
• UNNEST output struct subfields were pruned by the output's own access group instead of the input array's, which could prune subfields that were still needed. #76002
• The FE Iceberg manifest data-file cache could serve an incomplete file set, silently dropping a data file from scan planning and under-counting query results. #76215
• The Iceberg table metadata cache was not invalidated after an INSERT commit, so subsequent queries could miss the latest snapshot. #67230