PostgreSQL 15 deprecation announcement

We will continue supporting PostgreSQL 15 until June 2026. Closer to that time, we will announce the specific TimescaleDB version in which PostgreSQL 15 support will not be included going forward.

Features

• #8373 More precise estimates of row numbers for columnar storage based on Postgres statistics.
• #8581 Allow mixing Postgres and TimescaleDB options in ALTER TABLE SET.
• #8582 Make partition_column in CREATE TABLE WITH optional.
• #8588 Automatically create a columnstore policy when a hypertable with columnstore enabled is created via CREATE TABLE WITH statement.
• #8606 Add job history config parameters for maximum successes and failures to keep for each job.
• #8632 Remove ChunkDispatch custom node.
• #8637 Add INSERT support for direct compress.
• #8661 Allow ALTER TABLE ONLY to change reloptions to apply setting changes only to future chunks.
• #8703 Allow set-returning functions in continuous aggregates.
• #8734 Support direct compress when inserting into a chunk.
• #8741 Add support for unlogged hypertables.
• #8769 Remove continuous aggregate invalidation trigger.
• #8798 Enable UUIDv7 compression by default.
• #8804 Remove insert_blocker trigger.

Bugfixes

• #8561 Show warning when direct compress is skipped due to triggers or unique constraints.
• #8567 Do not require a job to have executed to show status.
• #8654 Fix approximate_row_count for compressed chunks.
• #8704 Fix direct DELETE on compressed chunk.
• #8728 Don't block dropping hypertables with other objects.
• #8735 Fix ColumnarScan for UNION queries.
• #8739 Fix cached utility statements.
• #8742 Potential internal program error when grouping by bool columns of a compressed hypertable.
• #8743 Modify schedule interval for job history pruning.
• #8746 Support show/drop chunks with UUIDv7 partitioning.
• #8753 Allow sorts over decompressed index scans for ChunkAppend.
• #8758 Improve error message on catalog version mismatch.
• #8774 Add GUC for WAL based invalidation of continuous aggregates.
• #8782 Stops sparse index from allowing multiple options.
• #8799 Set next_start for WITH clause compression policy.
• #8807 Only warn but not fail the compression if bloom filter indexes are configured but disabled with a GUC.

GUCs

• cagg_processing_wal_batch_size: Batch size when processing WAL entries.
• enable_cagg_wal_based_invalidation: Enable experimental invalidations for continuous aggregates using WAL.
• enable_direct_compress_insert: Enable direct compression during INSERT.
• enable_direct_compress_insert_client_sorted: Enable direct compress INSERT with presorted data.
• enable_direct_compress_insert_sort_batches: Enable batch sorting during direct compress INSERT.

Thanks

• @brandonpurcell-dev For highlighting issues with show_chunks() and UUIDv7 partitioning
• @moodgorning for reporting an issue with the timescaledb_information.job_stats view
• @ruideyllot for reporting set-returning functions not working in continuous aggregates
• @t-aistleitner for reporting an issue with utility statements in plpgsql functions