Advanced Changefeed Configuration

On this page

The following sections describe performance, settings, configurations, and details to tune changefeeds:

• Changefeed performance
• High durability delivery
• High throughput

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

• Kafka sinks
• Webhook sinks

Changefeed performance

New in v23.1: By default, changefeeds are integrated with elastic CPU, which helps to prevent changefeeds from affecting foreground traffic. For example, changefeed backfills and initial scans can be CPU-intensive. This integration will result in a cluster prioritizing SQL traffic over changefeeds. Since this may affect changefeed latency, you can monitor your cluster's admission control system on the Overload Dashboard and changefeed latency on the Changefeed Dashboard.

This is controlled by the following cluster settings, which are by default enabled:

changefeed.cpu.per_event_elastic_control.enabled
kvadmission.rangefeed_catchup_scan_elastic_control.enabled

For a more technical explanation of elastic CPU, refer to the Rubbing control theory on the Go scheduler blog post.

MuxRangefeed

MuxRangefeed is a subsystem that improves the performance of rangefeeds with scale. It significantly reduces the overhead of running rangefeeds. Without MuxRangefeed enabled the number of RPC streams is proportional with the number of ranges in a table. For example, a large table could have tens of thousands of ranges. With MuxRangefeed enabled, this proportion improves so that the number of RPC streams is relative to the number of nodes in a cluster.

You can enable its functionality with the changefeed.mux_rangefeed.enabled cluster setting.

Use the following workflow to enable MuxRangefeed:

1. Enable the cluster setting:

   SET CLUSTER SETTING changefeed.mux_rangefeed.enabled = true;
   

2. After enabling the setting, pause the changefeed:

   PAUSE JOB {job ID};
   

   You can use SHOW CHANGEFEED JOBS to retrieve the job ID.

3. Resume the changefeed for the cluster setting to take effect:

   RESUME JOB {job ID};
   

Latency in changefeeds

When you are running large workloads, changefeeds can encounter or cause latency in a cluster in the following ways:

• Changefeeds can have an impact on SQL latency in the cluster generally.
• Changefeeds can encounter latency in events emitting. This latency is the total time CockroachDB takes to:
  • Commit writes to the database.
  • Encode changefeed messages.
  • Deliver the message to the sink.

The following cluster settings reduce bursts of rangefeed work so that updates are paced steadily over time.

kv.closed_timestamp.target_duration

Default: 3s

kv.closed_timestamp.target_duration controls the target closed timestamp lag duration, which determines how far behind the current time CockroachDB will attempt to maintain the closed timestamp. For example, with the default value of 3s, if the current time is 12:30:00 then CockroachDB will attempt to keep the closed timestamp at 12:29:57 by possibly retrying or aborting ongoing writes that are below this time.

A changefeed aggregates checkpoints across all ranges, and once the timestamp on all the ranges advances, the changefeed can then checkpoint. In the context of changefeeds, kv.closed_timestamp.target_duration affects how old the checkpoints will be, which will determine the latency before changefeeds can consider the history of an event complete.

kv.rangefeed.closed_timestamp_refresh_interval

Default: 0s

This setting controls the interval at which closed timestamp updates are delivered to rangefeeds and in turn emitted as a changefeed checkpoint.

At the 0s interval default for this setting, every range with a rangefeed will emit a checkpoint event based on the kv.closed_timestamp.side_transport_interval value (default 200ms). Increasing the interval value for kv.rangefeed.closed_timestamp_refresh_interval will lengthen the delay between each checkpoint, which will increase the latency of changefeed checkpoints, but reduce the impact on SQL latency due to overload on the cluster.

If you are running changefeeds at a large scale and notice foreground SQL latency, we recommend increasing this setting.

As a result, adjusting kv.rangefeed.closed_timestamp_refresh_interval can affect changefeeds encountering latency and changefeeds causing foreground SQL latency. In clusters running large-scale workloads, it may be helpful to:

• Decrease the value for a lower changefeed emission latency — that is, how often a client can confirm that all relevant events up to a certain timestamp have been emitted.
• Increase the value to reduce the potential impact of changefeeds on SQL latency. This will lower the resource cost of changefeeds, which can be especially important for workloads with tables in the TB range of data.

It is important to note that a changefeed at default configuration does not checkpoint more often than once every 30 seconds. When you create a changefeed with CREATE CHANGEFEED, you can adjust this with the min_checkpoint_frequency option.

kv.closed_timestamp.side_transport_interval

Default: 200ms

The kv.closed_timestamp.side_transport_interval cluster setting controls how often the closed timestamp is updated. Although the closed timestamp is updated every 200ms by default, CockroachDB will emit an event across the rangefeed containing the closed timestamp value as per the cluster's kv.rangefeed.closed_timestamp_refresh_interval setting, which by default is 0s.

kv.closed_timestamp.side_transport_interval is helpful when ranges are inactive. The closed timestamp subsystem usually propagates closed timestamps via Raft commands. However, an idle range that does not see any writes does not receive any Raft commands, so it would stall. This setting is an efficient mechanism to broadcast closed timestamp updates for all idle ranges between nodes.

Adjusting kv.closed_timestamp.side_transport_interval will affect both follower reads and changefeeds. While you can use kv.closed_timestamp.side_transport_interval to tune the checkpointing interval, we recommend kv.rangefeed.closed_timestamp_refresh_interval if you are using follower reads.

kv.rangefeed.closed_timestamp_smear_interval

Default: 1ms

This setting provides a mechanism to pace the closed timestamp notifications to follower replicas. At the default, the closed timestamp smear interval makes rangefeed closed timestamp delivery less spiky, which can reduce its impact on foreground SQL query latency.

For example, if you have a large table, and one of the nodes in the cluster is hosting 6000 ranges from this table. Normally, the rangefeed system will wake up every kv.rangefeed.closed_timestamp_refresh_interval setting and will publish checkpoints for all 6000 ranges. In this scenario, the kv.rangefeed.closed_timestamp_smear_interval setting takes the kv.rangefeed.closed_timestamp_refresh_interval frequency and divides it into 1ms chunks. Instead of publishing checkpoints for all 6000 ranges, it will publish checkpoints for the calculated ranges every 1ms. This produces a more predictable and level load, rather than spiky, large bursts of workload.

Lagging ranges

New in v23.1.12: Use the changefeed.lagging_ranges metric to track the number of ranges that are behind in a changefeed. This is calculated based on the cluster settings:

• changefeed.lagging_ranges_threshold sets a duration from the present that determines the length of time a range is considered to be lagging behind, which will then track in the lagging_ranges metric. Note that ranges undergoing an initial scan for longer than the threshold duration are considered to be lagging. Starting a changefeed with an initial scan on a large table will likely increment the metric for each range in the table. As ranges complete the initial scan, the number of ranges lagging behind will decrease.
  • Default: 3m
• changefeed.lagging_ranges_polling_interval sets the interval rate for when lagging ranges are checked and the lagging_ranges metric is updated. Polling adds latency to the lagging_ranges metric being updated. For example, if a range falls behind by 3 minutes, the metric may not update until an additional minute afterward.
  • Default: 1m

Tuning for high durability delivery

When designing a system that relies on high durability message delivery—that is, not missing any message acknowledgement at the downstream sink—consider the following settings and configuration in this section:

• Pausing changefeeds and garbage collection
• Defining Kafka message acknowledgment
• Choosing changefeed sinks
• Defining schema change behavior

Before tuning these settings, we recommend reading details on our changefeed at-least-once-delivery guarantee.

Pausing changefeeds and garbage collection

By default, protected timestamps will protect changefeed data from garbage collection up to the time of the checkpoint. Protected timestamps will protect changefeed data from garbage collection if the downstream changefeed sink is unavailable until you either cancel the changefeed or the sink becomes available once again.

However, if the changefeed lags too far behind, the protected changes could lead to an accumulation of garbage. This could result in increased disk usage and degraded performance for some workloads.

For more detail on changefeeds and protected timestamps, refer to Garbage collection and changefeeds.

To balance protecting change data and prevent the over-accumulation of garbage, Cockroach Labs recommends creating a changefeed with options to define your protection duration and monitoring your changefeed for protected timestamp record collection.

Protecting change data on pause

Create changefeeds with the following options so that your changefeed protects data when it is paused:

• protect_data_from_gc_on_pause: to protect changes while the changefeed is paused until you resume the changefeed.
• on_error=pause: to pause the changefeed when it encounters an error. By default, changefeeds treat errors as retryable apart from some exceptions.
• New in v23.1: gc_protect_expires_after: to automatically expire the protected timestamp records that are older than your defined duration and cancel the changefeed job.

Monitoring protected timestamp records

New in v23.1: You can monitor changefeed jobs for protected timestamp usage. We recommend setting up monitoring for the following metrics:

• jobs.changefeed.protected_age_sec: Tracks the age of the oldest protected timestamp record protected by changefeed jobs. We recommend monitoring if protected_age_sec is greater than gc.ttlseconds. As protected_age_sec increases, garbage accumulation increases. Garbage collection will not progress on a table, database, or cluster if the protected timestamp record is present.
• jobs.changefeed.currently_paused: Tracks the number of changefeed jobs currently considered paused. Since paused changefeed jobs can accumulate garbage, it is important to monitor the number of paused changefeeds.
• jobs.changefeed.expired_pts_records: Tracks the number of expired protected timestamp records owned by changefeed jobs. You can monitor this metric in conjunction with the gc_protect_expires_after option.
• jobs.changefeed.protected_record_count: Tracks the number of protected timestamp records held by changefeed jobs.

Defining Kafka message acknowledgment

To determine what a successful write to Kafka is, you can configure the kafka_sink_config option. The 'RequiredAcks' field specifies what a successful write to Kafka is. CockroachDB guarantees at least once delivery of messages—the 'RequiredAcks' value defines the delivery.

For high durability delivery, Cockroach Labs recommends setting:

kafka_sink_config='{'RequiredAcks': 'ALL'}'

ALL provides the highest consistency level. A quorum of Kafka brokers that have committed the message must be reached before the leader can acknowledge the write.

Choosing changefeed sinks

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.

Defining schema change behavior

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 in this section:

• Setting the resolved option
• Batching and buffering messages
• Configuring file and message format
• Configuring for tables with many ranges
• Adjusting concurrent changefeed work

Setting the resolved option

When a changefeed emits a resolved message, it force flushes all outstanding messages that have buffered, which will diminish your changefeed's throughput while the flush completes. Therefore, if you are aiming for higher throughput, we suggest setting the duration higher (e.g., 10 minutes), or not using the resolved option.

If you are setting the resolved option when you are aiming for high throughput, you must also consider the min_checkpoint_frequency option, which defaults to 30s. This option controls how often nodes flush their progress to the coordinating changefeed node. As a result, resolved messages will not be emitted more frequently than the configured min_checkpoint_frequency. Set this option to at least as long as your resolved option duration.

Batching and buffering messages

• Batch messages to your sink:
  • For a Kafka sink, refer to the Flush parameter for the kafka_sink_config option.
  • For a cloud storage sink, use the file_size parameter to flush a file when it exceeds the specified size.
  • For a webhook sink, refer to the Flush parameter for the webhook_sink_config option.
• Set the changefeed.memory.per_changefeed_limit cluster setting to a higher limit to give more memory for buffering changefeed data. This setting influences how often the changefeed will flush buffered messages. This is useful during heavy traffic.

Configuring file and message format

• Use avro as the emitted message format option with Kafka sinks; JSON encoding can potentially create a slowdown.

Compression

• Use the compression option when you create a changefeed emitting data files to a cloud storage sink. For larger files, set compression to the zstd format.
• Use the snappy compression format to emit messages to a Kafka sink. If you're intending to do large batching for Kafka, use the lz4 compression format.

File size

To configure changefeeds emitting to cloud storage sinks for high throughput, you should consider:

• Increasing the file_size parameter to control the size of the files that the changefeed sends to the sink. The default is 16MB. To configure for high throughput, we recommend 32MB–128MB. Note that this is not a hard limit, and a changefeed will flush the file when it reaches the specified size.
• When you compress a file, it will contain many more events.
• File size is also dependent on what kind of data the changefeed job is writing. For example, large JSON blobs will quickly fill up the file_size value compared to small rows.
• When you change or increase file_size, ensure that you adjust the changefeed.memory.per_changefeed_limit cluster setting, which has a default of 512MiB. Buffering messages can quickly reach this limit if you have increased the file size.

Configuring for tables with many ranges

If you have a table with 10,000 or more ranges, you should consider increasing the following two cluster settings. We strongly recommend increasing these settings slowly. That is, increase the setting and then monitor its impact before adjusting further:

• kv.rangefeed.catchup_scan_concurrency: The number of catchups a rangefeed can execute concurrently. The default is 8.
• kv.rangefeed.concurrent_catchup_iterators: The number of rangefeed catchup iterators a store will allow concurrently before queuing. The default is 16.

Adjusting concurrent changefeed work

• 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.
• The kv.rangefeed.catchup_scan_iterator_optimization.enabled setting is on by default. This causes rangefeeds to 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

• Cluster Settings
• Changefeed Sinks
• CREATE CHANGEFEED

Was this helpful?