Version 5.4.1¶

Released on 2023-08-04.

Note

If you are upgrading a cluster, you must be running CrateDB 4.0.2 or higher before you upgrade to 5.4.1.

We recommend that you upgrade to the latest 5.3 release before moving to 5.4.1.

Before upgrading, you should back up your data.

Warning

A rolling upgrade from 5.3.x to 5.4.1 is not supported if your cluster contains view definitions. If you want to do a rolling upgrade, skip 5.4.1 and upgrade to 5.4.2.

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 5.4.0 release notes for a full list of changes in the 5.4 series.

Known Issues ¶

Version 5.0.0 introduced a regression which can cause some JOIN queries to return no results when the query optimizer re-orders the joined tables. As a workaround, users should apply the following session settings before running such queries to prevent the query optimizer from re-ordering them and therefore produce the correct results:

On CrateDB >= 5.4.0:
```
SET optimizer_reorder_hash_join = false
SET optimizer_reorder_nested_loop_join = false
```
On CrateDB < 5.4.0 (this will disable the hash-join algorithm which may lead to poor performance):
```
SET enable_hashjoin = false
```

Fixes ¶

Updated the bundled OpenJDK to 20.0.2+9
Allowed user to RESET cluster settings with archived. prefix.
Fixed an issue that could lead to a class_cast_exception when using a subscript expression on an aliased expression that shadows another column. For example:
```
SELECT a['b'] FROM (SELECT UNNEST(a) AS a FROM tbl) t;
```
Fixed an issue that caused SET SESSION statements to fail for settings that are both session settings and global settings, like statement_timeout.
Fixed an issue that caused a object_column = {} query to not match if the object column doesn’t have any child columns.
Fixed an issue that caused queries against views to evaluate the view definition with the current search path. This could cause issues if the search path diverged. The fix only applies to newly created views.
Fixed a Failed to validate IP error that occurred when using the any_value aggregation on a column of type IP.
Improved error message when passing settings with WITH clause on a CREATE SUBSCRIPTION statement, which is not currently supported.
Fixed an issue that resulted in hiding errors with the connection URL of CREATE SUBSCRIPTION, when executed from a non-superuser but with AL privileges.
Allowed trailing / without a database name for connection URL of CREATE SUBSCRIPTION.
Fixed a NullPointerException which could happen if using a cross join on a sub-query, where the sub-query was executed using a Fetch operator. An example query:
```
SELECT
  *
FROM
  (SELECT a FROM tbl1 ORDER BY b DESC LIMIT 1) i,
  tbl2
WHERE
  c >= 50;
```
Fixed a NullPointerException which was thrown, instead of using default no compression behavior, when compression parameter of COPY TO statement is set to null.

Fixed IndexOutOfBoundsException caused by an IS [NOT] NULL filter on a sub-column of an object or object array in a WHERE clause, e.g.

CREATE TABLE test (o1 ARRAY(OBJECT AS (col INT)), o2 OBJECT);
SELECT * FROM test WHERE o1[1]['col'] IS NULL;
=> IndexOutOfBoundsException[Index: 1 Size: 1]
SELECT * FROM test AS T WHERE T.o2['unknown_col'] IS NOT NULL;
=> IndexOutOfBoundsException[Index: 1 Size: 1]

Fixed an issue which caused INSERT INTO statements to skip generated expression validation for partitioned columns.
Fixed an issue which caused arrays in IGNORED objects to be converted to nulls.
Fixed an issue that caused LIKE/ILIKE operators on INDEX OFF columns to return empty results.

Fixed an issue that caused LIKE ANY/ILIKE ANY operators to accept only left arguments as patterns. Both arguments can now be patterns, e.g.

SELECT 'a' LIKE ANY(['a%']);
+-------+
| false |
+-------+
| FALSE | -- true will be returned after the fix
+-------+

Fixed an issue which caused INSERT INTO ... SELECT ... statements to leave behind empty partitions if NULL or CHECK constraint on partitioned by column failed.
Fixed an issue which caused errors on querying information_schema tables when the query of a VIEW is erroneous due to changes made to the underlying tables/views. Also, added a comment to view definition in pg_catalog.pg_views and information_schema.views tables to denote that a VIEW’s query is erroneous.

Fixed SQLParseException caused by querying an unknown key from an object column of a table that is aliased and with the session setting error_on_unknown_object_key, set to false, e.g.

CREATE TABLE test (o OBJECT);
SELECT T.o['unknown'] from (SELECT * FROM test) AS T;
=> SQLParseException[Couldn't create execution plan from logical plan because of: Couldn't find o['unknown'] in SourceSymbols{inputs={}, nonDeterministicFunctions={}}

Version 5.4.1¶

Known Issues¶

Fixes¶

Known Issues ¶

Fixes ¶