Search docs.molo17.com

    Microsoft SQL Server CDC logo MS SQL Server CDC agent changelog

    Release notes

    2.3.2.2

    Released: November 13, 2026

    • Views are listed by discovery and can be replicated as snapshot-only entities, left out of the Change Tracking and CDC setup, also when the entity is created through MCP;

    • Resolving a restart point in time works on SQL Server 2008 to 2014 instead of failing on AT TIME ZONE;

    • The "Trim text columns" option uses LTRIM and RTRIM instead of TRIM, so entities with it no longer fail every read on SQL Server 2008 to 2016;

    • The Query Studio CREATE TABLE is built without STRING_AGG, so it works on SQL Server before 2017 and lists the columns in table order;

    2.3.2.0

    Released: November 10, 2026

    • GSSD-1487: CDC changes are read in LSN order (__$start_lsn, __$command_id, __$seqval, __$operation) instead of the order the query plan produced, which delivered an insert after the updates that followed it and could move the read position past rows not read yet;

    • GSSD-1487: the ordered read follows the clustered index of the change table, so a window of 1.2 million changes returns its first row after 3 ms instead of 2.2 s, with no sort and no 371 MB memory grant;

    • A table SQL Server captures is recognised by name at sync start, so an entity is no longer refused with Change data capture not enabled on a table that is captured;

    • GSSD-1492: with Enable automatic Change Data Capture activation off (the default), the tables setup warns about the tables without capture and gives the statements to run by hand, instead of saying nothing until the entity failed at sync start;

    2.3.1.3

    Released: October 29, 2026

    • CDC: a read position older than the capture instance minimum resumes at the minimum instead of one LSN past it, which skipped the oldest rows the cleanup had kept;

    • CDC: log sequence numbers are compared as unsigned bytes, so a pipeline with two CDC targets no longer resumes from the wrong saved position;

    2.3.1.2

    Released: October 28, 2026

    • A time in the hour the agent’s time zone skips at the daylight saving change is replicated as the wall clock the source holds instead of an hour later, and a DATETIMEOFFSET column keeps its instant;

    2.3.1.1

    Released: October 27, 2026

    • Target: an object field from a NoSQL source is written as its JSON text instead of {name=Ada, age=36}, and an array that holds a map or a nested array is written as a JSON array instead of [Ljava.lang.Object;@…​;

    2.3.1.0

    Released: October 25, 2026

    • Target: snapshot pages after the first are written through one bulk copy kept open across pages, for much faster snapshots on wide tables and high-latency links. New settings directCopyChunkSizeMb (default 64), directCopyChunkSeconds (default 10) and directCopyChunkIdleSeconds (default 2); a size of 0 restores the per-page copy;

    • Target: each bulk copy is sent in one batch, a write stopped during the staging copy no longer leaves rows that a later write merges twice, and date-time columns of created tables are datetime2 (precision capped at 7);

    2.3.0.2

    Released: October 19, 2026

    • Bulk operations: snapshot rows are copied straight into the target table, without a staging table, when recursion protection is off and the target table has a primary key or a unique index on the entity’s key columns (an index created with IGNORE_DUP_KEY does not count). This includes the snapshot rows that sources such as PostgreSQL send as updates. A key the target already holds rolls the chunk back and writes it through the staging table, which the entity then keeps for the rest of the snapshot. Target triggers fire as they did through the staging table. Other tables use the staging table as before;

    • Bulk operations: bulk copy no longer fails with a ClassCastException on tables with real columns;

    • Bulk operations: bulk copy keeps the full precision of datetime2 values, up to the precision of the column, instead of cutting them to milliseconds;

    • Bulk operations: bulk copy keeps non-Latin characters in xml columns instead of turning them into question marks;

    • Multi-dimensional arrays are written as text like [1, 2], [3, 4] instead of a Java object name, in bulk and batch writes, and a binary value inside an array is written as 0x followed by hexadecimal;

    • Batch writes and bulk staging of wide tables are faster: each column’s value is looked up once per row instead of once per value;

    2.3.0.1

    Released: October 15, 2026

    • Bulk operations: an insert and a delete, or two updates, of a row keyed by binary or array values in one batch are merged instead of leaving the row on the target or failing with a duplicate key;

    • The agent hands its pooled database back when its pipeline is deleted or detached and when its initialization fails. Before, the next pipeline writing to the same database could fail in a loop with NoClassDefFoundError;

    2.3.0

    Released: October 13, 2026

    • Agent compatibility upgrade to Gluesync 2.3;

    • A migration rewrites the former disableAutomaticDatabaseCDCActivation flag so upgrades are not blocked; snapshot-only entities and unchanged saves do not re-fire the warning;

    • Added opt-in automatic Change Data Capture activation ("Enable automatic Change Data Capture / Change Tracking activation", off by default): the agent enables capture on the database, the tables and the recursion-protection sidecar itself. Left off, setup posts a warning with the statements to run by hand; turned on, a failed enable fails setup with the statement and the server error;

    • Columns are discovered in their table order instead of the order the query plan happened to produce;

    • Discovery returns schemas and tables in alphabetical order, so Query Studio pages no longer skip or repeat objects;