Advanced Changefeed Configuration

On this page Carat arrow pointing down
Warning:

The configurations and settings explained on this page will significantly impact a changefeed's behavior.

The following sections describe settings, configurations, and details to tune changefeeds for these use cases:

Some specific options for the kafka_sink_config and webhook_sink_config parameters are discussed on this page. However, for more information on more specific tuning for Kafka and Webhook sinks see the following pages:

Tuning for high durability delivery

When designing a system that relies on high durability of message delivery — that is, not missing any message acknowledgement at the downstream sink — consider the following settings and configuration. Before tuning these settings we recommend reading details on our changefeed at-least-once-delivery guarantee.

  • Increase the number of seconds before garbage collection with the gc.ttlseconds setting to provide a higher recoverability window for data if a changefeed fails. For example, if a sink is unavailable, changes are queued until the sink is available again. While the sink is unavailable, changes will be retried until the garbage collection window is reached and then the data is removed.
    • You can also use the protect_data_from_gc_on_pause option in combination with on_error=pause to explicitly pause a changefeed on a non-retryable error (instead of going into a failure state) and to then protect the changes from garbage collection.
  • Determine what a successful write to Kafka is with the kafka_sink_config: {"RequiredAcks": "ALL"} option, which provides the highest consistency level.
  • Use Kafka or cloud storage sinks when tuning for high durability delivery in changefeeds. Both Kafka and cloud storage sinks offer built-in advanced protocols, whereas the webhook sink, while flexible, requires an understanding of how messages are acknowledged and committed by the particular system used for the webhook in order to ensure the durability of message delivery.
  • Ensure that data is ingested downstream in its new format after a schema change by using the schema_change_events and schema_schange_policy options. For example, setting schema_change_events=column_changes and schema_change_policy=stop will trigger an error to the cockroach.log file on a schema change and the changefeed to fail.

Tuning for high throughput

When designing a system that needs to emit a lot of changefeed messages, whether it be steady traffic or a burst in traffic, consider the following settings and configuration:

  • Avoid using the resolved option or set this to a higher duration. This will help to reduce emitted messages.
  • Batch messages to your sink. See the Flush parameter for the kafka_sink_config option. When using cloud storage sinks, use the file_size parameter to flush a file when it exceeds the specified size.
  • Set the changefeed.memory.per_changefeed_limit cluster setting to a higher limit to give more memory for buffering for a changefeed. This is useful in situations of heavy traffic.
  • Use avro as the emitted message format option with Kafka sinks; JSON encoding can potentially create a slowdown.
  • Use the compression option in cloud storage sinks with JSON to compress the changefeed data files.
  • Increase the changefeed.backfill.concurrent_scan_requests setting, which controls the number of concurrent scan requests per node issued during a backfill event. The default behavior, when this setting is at 0, is that the number of scan requests will be 3 times the number of nodes in the cluster (to a maximum of 100). While increasing this number will allow for higher throughput, it will increase the cluster load overall, including CPU and IO usage.
  • New in v21.2: Enable the kv.rangefeed.catchup_scan_iterator_optimization.enabled setting to have rangefeeds use time-bound iterators for catch-up scans when possible. Catch-up scans are run for each rangefeed request. This setting improves the performance of changefeeds during some range-split operations.

See also


Yes No
On this page

Yes No