CDC-to-Kafka 2.1.1

Fixes a few minor bugs:

1. Progress deletion tombstone message type not recognized when collecting stats
2. Upon capture instance upgrades, exceptions were sometimes raised on LSNs being set to None
3. Upon capture instance upgrades, tables that had no data in their capture instances would lead to the process restart that picks up the new instances being delayed indefinitely

CDC-to-Kafka 2.1.0

Two minor improvements:

• The process now specifies a different default Avro schema compatibility level for the "unified topic" messages. Previously, FORWARD compatibility was the default for all schema registry subjects created by this process. However, the unified topic's schema will often see a new record type added to its union type which represents the schemas of all tracked tables (i.e. whenever a new table starts being tracked), and adding a new member to an Avro union type is not a forward-compatible change. So we fall back to BACKWARD because it's about the best we can do. Single-table topics continue to default to FORWARD compatibility.
• Prevents altering the compatibility level of a preexisting schema registry subject. The default compatibility levels are applied only if/when this process is creating a new subject in the registry.

CDC-to-Kafka 2.0.1

This is a minor bugfix which upgrade fields in the Avro schema of progress-tracking messages from int to long, to handle cases where change data topics accumulate more than 2^31 messages.

CDC-to-Kafka 2.0.0

This is a major refactor providing several new features:

• Production to "unified" topics that can contain properly-ordered changes across all or a selected subset of tables (i.e. a "DIY SQL Server tlog")
• Enhanced detection and handling of capture instance changes, with schema evolution and restart-on-change capability
• Improved data validation scripts
• General performance improvements, particularly around speed of table snapshots

CDC-to-Kafka 1.2.1

• Fixes a performance issue with snapshot queries against tables with composite primary keys

CDC-to-Kafka 1.2.0

Adds the ability to specify a truncation length on specific fields, to work around large SQL rows causing Kafka producer errors due to oversized messages.

CDC-to-Kafka 1.1.2

• Improves performance of change table queries that have different sets of clustered index columns

CDC-to-Kafka 1.1.1

• Resolves "sleep length must be non-negative" errors
• Drops the confusing and relatively useless cdc_to_kafka_process_lag_sec metric (use the improved e2e_latency_* metrics instead!)

CDC-to-Kafka 1.1.0

Improves process metrics reporting

CDC-to-Kafka 1.0.0

Initial release