Version 5.7.2

Released on 2024-06-12.

Note

If you are upgrading a cluster, you must be running CrateDB 4.0.2 or higher before you upgrade to 5.7.2.

We recommend that you upgrade to the latest 5.6 release before moving to 5.7.2.

A rolling upgrade from 5.6.x to 5.7.2 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 5.7.0 release notes for a full list of changes in the 5.7 series.

Security Fixes

  • Disabled client-initiated renegotiation for TLS by default. This helps prevent Denial of Service (DoS) attacks.

Fixes

  • Fixed an issue that prevented users with DDL privileges on table to execute ALTER TABLE t REROUTE… statements.

  • Fixed an issue that could lead to requests getting stuck when trying to download a blob via HTTPS.

  • Fixed an issue leading to slow query processing during the analysis phase, when the WHERE clause of a query contains columns of a PRIMARY KEY and combines them using complex logical expressions, e.g.:

    SELECT * FROM t WHERE pk_col1 = ? AND pk_col2 = ? OR
                      pk_col1 = ? AND pk_col2 = ? OR
                      ...

  • Fixed an issue leading to a UnsupportedFeatureException when using a correlated sub-query in a case function as part of a select statement where some of its outputs weren’t used in the outer-query.

  • Fixed an issue leading to a ArrayIndexOutOfBoundsException instead of a user friendly error message when the WHERE clause of a query contains all columns of a PRIMARY KEY, uses parameters for them, and binds less actual values than the required, e.g.:

    SELECT * FROM t WHERE pk_col1 = ? AND pk_col2 = ? AND pk_col3 = ?

    and less than 3 values are provided.

  • Added memory accounting for multi-phase execution to prevent out-of-memory errors caused by sub-queries such as SELECT * FROM t1 WHERE id IN (SELECT id FROM t2) or lookup-joins with large intermediate results.

  • Fixed an issue that caused WHERE clause to fail to filter rows when the clause contained CONCAT, CURRENT_SETTING, PG_GET_FUNCTION_RESULT, or PG_ENCODING_TO_CHAR scalar functions under NOT operator.

  • Fixed an issue leading to restoring data into wrong tables when restoring a snapshot with partitioned tables using one of the renaming parameters schema_rename_pattern, schema_rename_replacement, table_rename_pattern or table_rename_replacement. Renamed table was created but used to be empty.

  • Fixed an issue that prevented users with DQL privilege on a view to be able to query the view when foreign tables are referenced by the view.

  • Fixed an issue leading to a ArrayIndexOutOfBoundsException when using a correlated sub-query where some of its outputs weren’t used in a parent query.

  • Fixed an issue that could lead to out-of-memory errors when executing lead or lag window functions on a large table.

  • Fixed an issue that could lead to errors when loading data from geo_point fields inside nested arrays.

  • Fixed an issue that caused SQLParseException when the results returned from foreign tables could not be casted to CrateDB specific data types.

  • Fixed an issue that allowed dropped users to run queries if active user was set by SET SESSION AUTHORIZATION and then dropped.

  • Fixed an issue leading to authentication errors when Host-Based Authentication was enabled and had entries matching multiple authentication methods simultaneously, e.g.:

    auth.host_based.config.1.method=password
auth.host_based.config.2.method=jwt
auth.host_based.config.2.protocol=http

    This example configuration were rejecting authentication via JWT as only the first entry was checked.