Version 5.6.5

Released on 2024-05-02.


If you are upgrading a cluster, you must be running CrateDB 4.0.2 or higher before you upgrade to 5.6.5.

We recommend that you upgrade to the latest 5.5 release before moving to 5.6.5.

A rolling upgrade from 5.5.x to 5.6.5 is supported.

Before upgrading, you should back up your data.


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 5.6.0 release notes for a full list of changes in the 5.6 series.


  • Fixed an issue that could result in completed jobs to not be marked with an ended timestamp and be moved to sys.jobs_log, but stay “forever” (until node restart) in Jobs table.

  • Fixed an issue that allowed creating tables named _all, which is a reversed name that has special effects - like expanding to all table names. Because of these special semantics, statements like DROP TABLE _all would delete all tables instead of only _all.

  • Fixed an issue that could prevent a RESTORE SNAPSHOT from restoring a partition if using the TABLE PARTITION (...) clause to restore an individual partition of a table with more than one partition column.

  • Fixed an issue that allowed INSERT INTO statements to create more shards than allowed by cluster.max_shards_per_node if the statement resulted in the creation of many partitions at once.

  • Fixed an issue that resulted on the garbage collection logging environment variables being ignored in Windows.

  • Fixed an issue that caused assigning a wrong number of replicas to closed tables after cluster was re-balanced (for example by adding a node), if tables were created with number_of_replicas specified as a range.

  • Fixed an issue that caused queries on partitioned tables via prepared statements to return invalid results if the partitions of the table changed after preparing the statements.

  • Fixed an issue that caused prepared statements to cache values of non-deterministic functions and use the same value for each execution.

  • Improved ingestion performance for partitioned tables where no primary keys are defined by using an append only optimized ingestion strategy. The performance improvement increases with the size of each shard.