Version 5.5.1

Released on 2023-11-30.

Note

If you are upgrading a cluster, you must be running CrateDB 4.0.2 or higher before you upgrade to 5.5.1.

We recommend that you upgrade to the latest 5.4 release before moving to 5.5.1.

A rolling upgrade from 5.4.x to 5.5.1 is supported. Before upgrading, you should back up your data.

Warning

Tables that were created before CrateDB 4.x will not function with 5.x and must be recreated before moving to 5.x.x.

You can recreate tables using COPY TO and COPY FROM or by inserting the data into a new table.

Table of contents

See the Version 5.5.0 release notes for a full list of changes in the 5.5 series.

Fixes

• Fixed an issue that caused queries with a NOT expression in the WHERE clause to fail evaluating NULL correctly.

• Fixed an issue that caused the value for generated primary key columns to evaluate to NULL in INSERT INTO .. ON CONFLICT statements if the column wasn’t part of the target column list.

• Creating a table that uses a table-function as part of a default expression or generated expression now results in an error on table creation, instead of never inserting records due to runtime failures.

• Improved the error message when using COPY FROM with wait_for_completion=false and RETURN SUMMARY. It now reports that the combination is not supported instead of running into a ClassCastException.

• Fixed an issue that caused queries with a NOT (a AND b) expression in the WHERE clause to not evaluate correctly with NULL values.

• Fixed an issue that caused queries with a NOT or != on a CASE expression containing a nullable column to exclude NULL entries.

• Re-added jcmd to the bundled JDK distribution.

• Return meaningful error when trying to drop a column which itself or its sub-columns participates in a table level constraint with other columns. For example, we cannot drop column col_to_drop from the following tables:

  CREATE TABLE t1 (i int, col_to_drop int,
                   CHECK (col_to_drop + i > 0))
  CREATE TABLE t2 (i int, col_to_drop object AS (subcol_a int),
                   CHECK (col_to_drop['subcol_a'] + i > 0))
  

• Fixed an issue that caused a ColumnUnknownException to be thrown when attempting to drop a column with a CHECK. All relevant constraints are dropped together with the column.

• Fixed an issue that caused the hash-join operator generate invalid hashes which lead to a broken join operation when there were more than two relations involved e.g.:

  SELECT * FROM t1, t2, t3 WHERE t3.c = t1.a AND t3.c = t2.b AND t1.a = t2.b;
  

  would generate the logical plan:

  HashJoin[(t3.c = t2.b AND t1.a = t2.b)]
    ├ HashJoin[(t3.c = t1.a)]
    │  ├ Collect[doc.t3 | [c] | true]
    │  └ Collect[doc.t1 | [a] | true]
    └ Collect[doc.t2 | [b] | true]
  

  The hash-symbol generation for the join t3.c = t2.b AND t1.a = t2.b was broken and would not join the data.

• Fixed a regression introduced in 5.5.0 that caused the loss of filter conditions in nested-joins when the query was optimized by the rule optimizer_move_filter_beneath_join.

• Fixed a regression introduced in 5.5.0 that caused empty string to be accepted as a valid column name while table creation.