Version 4.4.0

Released on 2021-01-19.

Note

If you are upgrading a cluster, you must be running CrateDB 4.0.2 or higher before you upgrade to 4.4.0.

We recommend that you upgrade to the latest 4.3 release before moving to 4.4.0.

A rolling upgrade from 4.3.x to 4.4.0 is supported.

Before upgrading, you should back up your data.

Table of Contents

Deprecations

  • The settings discovery.zen.publish_timeout, discovery.zen.commit_timeout, discovery.zen.no_master_block, discovery.zen.publish_diff.enable have been marked as deprecated and will be removed in a future version.

Changes

Performance improvements

  • Improved the performance of queries on the sys.health table.
  • Added support for using the optimized primary key lookup plan if additional filters are combined via AND operators.
  • Improved the performance of queries on the sys.allocations table in cases where there are filters restricting the result set or if only a sub-set of the columns is selected.
  • Improved the performance for queries which select a subset of the columns available in a wide table.

SQL and PostgreSQL compatibility improvements

  • Added arithmetic operation support for the numeric type.
  • Added support for the numeric data type and allow the sum aggregation on the numeric type. Note that the storage of the numeric data type is not supported.
  • Extended the RowDescription message that can be sent while communicating with PostgreSQL clients to include a table_oid and a attr_num based on the values that are also exposed via the pg_catalog.pg_attribute table. This improves compatibility with clients which make use of these attributes.
  • Changed the format_type function to use the PostgreSQL compatible type name notation with [] suffixes for arrays, instead of _array.
  • Added the delimiter option for COPY FROM CSV files. The option is used to specify the character that separates columns within a row.
  • Added the empty_string_as_null option for COPY FROM CSV files. If the option is enabled, all column’s values represented by an empty string, including a quoted empty string, are set to NULL.

Administration and Operations

New scalar and window functions

  • Added the to_char scalar function for timestamp and interval argument data types.
  • Added the split_part scalar function.
  • Added the dense_rank window function, which is available as an enterprise feature.
  • Added the rank window function, which is available as an enterprise feature.

Feedback

How helpful was this page?