[RELEASE] ScyllaDB Enterprise 2021.1.17

The ScyllaDB team announces ScyllaDB Enterprise 2021.1.17, a bug-fix production-ready ScyllaDB Enterprise patch release for ScyllaDB Enterprise 2021.1.

Note that ScyllaDB Enterprise 2022.1 LTS is the latest stable branch and you are encouraged to upgrade in coordination with the ScyllaDB support team.

Related Links

The following issues are fixed in this release (with an open source reference):

  • Stability: Time window compaction strategy (TWCS) major compactions will now serialize behind other compactions, in order to to include all sstables in the compaction input set. #9553
  • Stability: Mismatched types for base and view columns (leftovers from #8666 fix)
  • Stability: compaction: Make compaction_manager stop more robust #11711
  • Stability: materialized view doesn’t work if there is a “to” column #9450
  • Stability: unpop partition header of spent partition in multishard mutation query. With the fix, Scylla does not resume an already exhausted partition at the start of generating a new page. #9482
  • In Time Window Compaction Strategy compactions, fully expired sstables will be compacted separately, since that can be done by just dropping them. #9533
  • Workload Prioritization improvements and bug fix:
    • Stability: removing a service level during an sstable load could lead to reading deleted memory and exit
    • Stability: some requests can ‘leak’ into the default service level just after authentication
  • Stability: use-after-free in db/view row_locker #12168
  • Stability: replace nodes are (at times) left in system.peers table - they need to be removed #8729
  • Stability: segmentation fault in cases where the base table schema change while MV schema is cached #10026, #11542
  • Stability on AWS EC2: From time to time Scylla reboot failed with “create *.Ec2Snitch: std::system_error (error system:110, Connection timed out)” error. #10250
  • Stability: UndefinedBehavior in repair_reader::make_reader #10882
  • Stability: The view builder is responsible for building materialized views when a view is created and after repair. To control its memory consumption, it reads base table sstables in chunks. Due to a bug, it could forget the partition tombstone or an open range tombstone when moving between chunks, leading to a discrepancy between the base table data and the view data. This is now fixed. #11668