Version 5.3.1

Released on 2023-04-28.

Warning

CrateDB 5.3.x versions up to 5.3.3 (excluding) contain a critical bug which can lead to data corruption/loss when using a column definition with a number data type and disabled index (INDEX OFF). It is not recommended to use those versions, use CrateDB >= 5.3.3 instead.

Note

If you are upgrading a cluster, you must be running CrateDB 4.0.2 or higher before you upgrade to 5.3.1.

We recommend that you upgrade to the latest 5.2 release before moving to 5.3.1.

A rolling upgrade from 5.3.x to 5.3.1 is supported. For upgrades from 5.2.x, see the warning below. Before upgrading, you should back up your data.

Warning

Due to a bug in the replication layer, rolling upgrades from 5.2.x to 5.3.1 with ongoing write traffic can lead to corrupted shards and in worse case, data loss. We recommend that you stop all write traffic before upgrading and/or perform a full cluster restart.

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 5.3.0 release notes for a full list of changes in the 5.3 series.

Fixes

  • Fixed a performance regression introduced in 5.2.3 which led to filters on object columns resulting in a table scan if used with views or virtual tables. See #14015 for details.

  • Fixed an issue that caused geo_shape_array IS NULL expressions to fail with an IllegalStateException.

  • Fixed an issue that caused the actual cast/type conversion error to be hidden when it failed for a sub-column of an object column, when using a client statement with parameters i.e (python).:

    CREATE TABLE a (b OBJECT(DYNAMIC) AS (c REAL));
# create a connection and a cursor and then:
cursor.execute("INSERT INTO a VALUES (?)", [({"c": True},)])

  • Fixed a regression that caused the -h option in bin/crate to fail with an Error parsing arguments! error.

  • Added a null_or_empty scalar function that can be used as a faster alternative to IS NULL if it’s acceptable to match on empty objects. This makes it possible to mitigate a performance regression introduced in 5.0.3 and 5.1.1

  • Fixed an issue that led to NullPointerException when trying to query an OBJECT field with no values, using the NOT operator, e.g.:

    CREATE TABLE test (obj OBJECT(DYNAMIC)); -- no data
SELECT myobj FROM test WHERE (obj::TEXT) NOT LIKE '%value%';

  • Fixed an issue in the PostgreSQL wire protocol implementation that could lead to ClientInterrupted errors with some clients. An example client is pg-cursor.

  • Fixed an issue that allowed creating columns with names conflicting with subscript pattern, such as "a[1]", a subscript expression enclosed in double quotes.

  • Fixed an issue that caused SQLParseException when quoted subscript expressions contained quotes. An example would be querying an array with the name containing quotes like SELECT "arr""[1]";.

  • Fixed an issue that caused ALTER TABLE ADD COLUMN statement to assign PRIMARY KEY to wrong columns, when adding multiple primary key columns, having none-primary columns in-between.

  • Fixed an issue that caused ALTER TABLE ADD COLUMN statement to assign a wrong type to ARRAY(TEXT) column and create a TEXT column instead if column has a FULLTEXT index.

  • Fixed an issue that prevented assigning default expression to ARRAY columns.

  • Reverted base image change in the Docker image as it broke downstream components which rely on some of the bundled tools like rev.