LogMiner Connector

1. Oracle Active DataGuard support for Oracle Database settings check utility
2. Fix for Oracle DataGuard when V$STANDBY_LOG does not contain rows
3. Fix ORA-310/ORA-334 under heavy RDBMS load
4. New parameters to support pseudo columns - a2.pseudocolumn.ora_rowscn, a2.pseudocolumn.ora_commitscn, a2.pseudocolumn.ora_rowts, & a2.pseudocolumn.ora_operation. For more information please read KAFKA-CONNECT.md
5. New parameters to support audit pseudo columns: a2.pseudocolumn.ora_username, a2.pseudocolumn.ora_osusername, a2.pseudocolumn.ora_hostname, a2.pseudocolumn.ora_audit_session_id, a2.pseudocolumn.ora_session_info, & a2.pseudocolumn.ora_client_id. For more information please read KAFKA-CONNECT.md

Sink Connector

New parameters: a2.table.mapper, a2.table.name.prefix, and a2.table.name.suffix

Simplification of configuration for Oracle Active DataGuard - now the same configuration is used for Oracle Active DataGuard as for a primary database

###LogMiner Connector

1. New parameter - a2.stop.on.ora.1284 to manage the connector behavior on ORA-1284. For more information please read KAFKA-CONNECT.md
2. Checking the number of non-zero columns returned from a redo record for greater reliability.
3. Handling of partial rollback records in RDBMS 19.13 i.e. when redo record with ROLLBACK=1 is before redo record with ROLLBACK=0
4. Processing of DELETE operation for tables ROWID pseudo key
5. New parameter - a2.print.unable.to.delete.warning to manage the connector output in log for DELETE operations over table's without PK. For more information please read KAFKA-CONNECT.md
6. New parameter - a2.schema.name.mapper to manage schema names generation. For more information please read KAFKA-CONNECT.md

###Docker image Rehost Confluent schema registry clients (Avro/Protobuf/JSON Schema) and bump version to 7.5.3

2.2.0 (MAR-2024)

LogMiner Connector

1. Enhanced handling of partial rollback redo records (ROLLBACK=1). For additional information about these redo records please read ROLLBACK INTERNALS starting with the sentence "The interesting thing is with partial rollback."
2. New parameter a2.topic.mapper to manage the name of the Kafka topic to which data will be sent. For more information please read KAFKA-CONNECT.md
3. Oracle Database settings check utility

Sink Connector

1. Connector classes are re-factored and the Sink Connector itself renamed from solutions.a2.cdc.oracle.OraCdcJdbcSinkConnector to solutions.a2.kafka.sink.JdbcSinkConnector
2. New parameter - a2.table.mapper to manage the table in which to sink the data.

2.1.0 (FEB-2024)

ServiceLoader manifest files, for more information please read KIP-898: Modernize Connect plugin discovery

LogMiner Connector

1. Now oracdc now also checks for first available SCN in V$LOG
2. Reducing the output about scale differences between redo and dictionary
3. Separate first available SCN detection for primary and standby

New parameters

a2.incomplete.redo.tolerance - to manage connector behavior when processing an incomplete redo record. For more information please read KAFKA-CONNECT.md a2.print.all.online.scn.ranges - to control output when processing online redo logs. For more information please read KAFKA-CONNECT.md a2.log.miner.reconnect.ms - to manage reconnect interval for LogMiner for Unix/Linux. For more information please read KAFKA-CONNECT.md a2.pk.type - to manage behavior when choosing key fields in schema for table. For more information please read KAFKA-CONNECT.md a2.use.rowid.as.key - to manage behavior when the table does not have appropriate PK/unique columns for key fields. For more information please read KAFKA-CONNECT.md a2.use.all.columns.on.delete - to manage behavior when reading and processing a redo record for DELETE. For more information please read KAFKA-CONNECT.md

Sink Connector

1. fix for SQL statements creation when first statement in topic is "incomplete" (delete operation for instance)
2. add exponential back-off for sink getConnection()
3. support for key-less schemas
4. PostgreSQL: support for implicitly defined primary keys

Online redo logs processing: Online redo logs are processed when parameter a2.process.online.redo.logs is set to true (Default - false). To control the lag between data processing in Oracle, the parameter a2.scn.query.interval.ms is used, which sets the lag in milliseconds for processing data in online logs. This expands the range of connector tasks and makes its use possible where minimal and managed latency is required.

default values: Column default values are now part of table schema

19c enhancements:

1. HEX('59') (and some other single byte values) for DATE/TIMESTAMP/TIMESTAMPTZ are treated as NULL
2. HEX('787b0b06113b0d')/HEX('787b0b0612013a')/HEX('787b0b0612090c')/etc (2.109041558E-115,2.1090416E-115,2.109041608E-115) for NUMBER(N)/NUMBER(P,S) are treated as NULL information about such values is not printed by default in the log; to print messages you need to set the parameter a2.print.invalid.hex.value.warning value to true

solution for incomplete redo information: Solution for problem described in LogMiner REDO_SQL missing WHERE clause and LogMiner Redo SQL w/o WHERE-clause

Support for INTERVALYM/INTERVALDS TIMESTAMP enhancements SDU hint in log

• New a2.pk.string.length parameter for Sink Connector and other Sink Connector enhancement
• New a2.transaction.implementation parameter for LogMiner Source Connector: when set to ChronicleQueue (default) oracdc uses Chronicle Queue to store information about SQL statements in Oracle transaction and uses off-heap memory and needs disk space to store memory mapped files; when set to ArrayList oracdc uses ArrayList to store information about SQL statements in Oracle transaction and uses JVM heap (no disk space needed).
• Fix for ORA-17002 while querying data dictionary
• Better handling for SQLRecoverableException while querying data dictionary

fix unhandled ORA-17410 running 12c on Windows and strict checks for supplemental logging settings

New a2.schema.type=single - schema type to store all columns from database row in one message with just value schema