Adjust the definition of min_checkpoint_frequency cdc

src/current/v23.1/create-changefeed.md

@@ -170,7 +170,7 @@ Option | Value | Description
 <span class="version-tag">New in v23.1:</span> <a name="key-column"></a>`key_column` | `'column'` | Override the key used in [message metadata]({% link {{ page.version.version }}/changefeed-messages.md %}). This changes the key hashed to determine downstream partitions. In sinks that support partitioning by message, CockroachDB uses the [32-bit FNV-1a](https://wikipedia.org/wiki/Fowler%E2%80%93Noll%E2%80%93Vo_hash_function) hashing algorithm to determine which partition to send to.<br><br>**Note:** `key_column` does not preserve ordering of messages from CockroachDB to the downstream sink, therefore you must also include the [`unordered`](#unordered) option in your changefeed creation statement. It does not affect per-key [ordering guarantees]({% link {{ page.version.version }}/changefeed-messages.md %}#ordering-and-delivery-guarantees) or the output of [`key_in_value`](#key-in-value).<br><br>See the [Define a key to determine the changefeed sink partition](#define-a-key-to-determine-the-changefeed-sink-partition) example.
 <a name="key-in-value"></a>`key_in_value` | N/A | Add a primary key array to the emitted message. This makes the [primary key]({% link {{ page.version.version }}/primary-key.md %}) of a deleted row recoverable in sinks where each message has a value but not a key (most have a key and value in each message). `key_in_value` is automatically used for [cloud storage sinks]({% link {{ page.version.version }}/changefeed-sinks.md %}#cloud-storage-sink), [webhook sinks]({% link {{ page.version.version }}/changefeed-sinks.md %}#webhook-sink), and [GC Pub/Sub sinks]({% link {{ page.version.version }}/changefeed-sinks.md %}#google-cloud-pub-sub).
 `metrics_label` | [`STRING`]({% link {{ page.version.version }}/string.md %}) | Define a metrics label to which the metrics for one or multiple changefeeds increment. All changefeeds also have their metrics aggregated.<br><br>The maximum length of a label is 128 bytes. There is a limit of 1024 unique labels.<br><br>`WITH metrics_label=label_name` <br><br>For more detail on usage and considerations, see [Using changefeed metrics labels]({% link {{ page.version.version }}/monitor-and-debug-changefeeds.md %}#using-changefeed-metrics-labels).
-<a name="min-checkpoint-frequency"></a>`min_checkpoint_frequency` | [Duration string](https://pkg.go.dev/time#ParseDuration) | Controls how often nodes flush their progress to the [coordinating changefeed node]({% link {{ page.version.version }}/how-does-an-enterprise-changefeed-work.md %}). Changefeeds will wait for at least the specified duration before a flush to the sink. This can help you control the flush frequency of higher latency sinks to achieve better throughput. However, more frequent checkpointing can increase CPU usage. If this is set to `0s`, a node will flush messages as long as the high-water mark has increased for the ranges that particular node is processing. If a changefeed is resumed, then `min_checkpoint_frequency` is the amount of time that changefeed will need to catch up. That is, it could emit [duplicate messages]({% link {{ page.version.version }}/changefeed-messages.md %}#duplicate-messages) during this time. <br><br>**Note:** [`resolved`](#resolved-option) messages will not be emitted more frequently than the configured `min_checkpoint_frequency` (but may be emitted less frequently). If you require `resolved` messages more frequently than `30s`, you must configure `min_checkpoint_frequency` to at least the desired `resolved` message frequency. For more details, refer to [Resolved message frequency]({% link {{ page.version.version }}/changefeed-messages.md %}#resolved-timestamp-frequency).<br><br>**Default:** `30s`
+<a name="min-checkpoint-frequency"></a>`min_checkpoint_frequency` | [Duration string](https://pkg.go.dev/time#ParseDuration) | Controls how often a node's changefeed [aggregator]({% link {{ page.version.version }}/how-does-an-enterprise-changefeed-work.md %}) will flush their progress to the [coordinating changefeed node]({% link {{ page.version.version }}/how-does-an-enterprise-changefeed-work.md %}). A node's changefeed aggregator will wait at least the specified duration between sending progress updates for the ranges it is watching to the coordinator. This can help you control the flush frequency of higher latency sinks to achieve better throughput. However, more frequent checkpointing can increase CPU usage. If this is set to `0s`, a node will flush messages as long as the high-water mark has increased for the ranges that particular node is processing. If a changefeed is resumed, then `min_checkpoint_frequency` is the amount of time that changefeed will need to catch up. That is, it could emit [duplicate messages]({% link {{ page.version.version }}/changefeed-messages.md %}#duplicate-messages) during this time. <br><br>**Note:** [`resolved`](#resolved-option) messages will not be emitted more frequently than the configured `min_checkpoint_frequency` (but may be emitted less frequently). If you require `resolved` messages more frequently than `30s`, you must configure `min_checkpoint_frequency` to at least the desired `resolved` message frequency. For more details, refer to [Resolved message frequency]({% link {{ page.version.version }}/changefeed-messages.md %}#resolved-timestamp-frequency).<br><br>**Default:** `30s`
 `mvcc_timestamp` | N/A |  Include the [MVCC]({% link {{ page.version.version }}/architecture/storage-layer.md %}#mvcc) timestamp for each emitted row in a changefeed. With the `mvcc_timestamp` option, each emitted row will always contain its MVCC timestamp, even during the changefeed's initial backfill.
 <a name="on-error"></a>`on_error` | `pause` / `fail` |  Use `on_error=pause` to pause the changefeed when encountering **non**-retryable errors. `on_error=pause` will pause the changefeed instead of sending it into a terminal failure state. **Note:** Retryable errors will continue to be retried with this option specified. <br><br>Use with [`protect_data_from_gc_on_pause`](#protect-pause) to protect changes from [garbage collection]({% link {{ page.version.version }}/configure-replication-zones.md %}#gc-ttlseconds).<br><br>If a changefeed with `on_error=pause` is running when a watched table is [truncated]({% link {{ page.version.version }}/truncate.md %}), the changefeed will pause but will not be able to resume reads from that table. Using [`ALTER CHANGEFEED`]({% link {{ page.version.version }}/alter-changefeed.md %}) to drop the table from the changefeed and then [resuming the job]({% link {{ page.version.version }}/resume-job.md %}) will work, but you cannot add the same table to the changefeed again. Instead, you will need to [create a new changefeed](#start-a-new-changefeed-where-another-ended) for that table.<br><br>Default: `on_error=fail`
 <a name="protect-pause"></a>`protect_data_from_gc_on_pause` | N/A |  When a [changefeed is paused]({% link {{ page.version.version }}/pause-job.md %}), ensure that the data needed to [resume the changefeed]({% link {{ page.version.version }}/resume-job.md %}) is not garbage collected. If `protect_data_from_gc_on_pause` is **unset**, pausing the changefeed will release the existing protected timestamp records. It is also important to note that pausing and adding `protect_data_from_gc_on_pause` to a changefeed will not protect data if the [garbage collection]({% link {{ page.version.version }}/configure-replication-zones.md %}#gc-ttlseconds) window has already passed. <br><br>Use with [`on_error=pause`](#on-error) to protect changes from garbage collection when encountering non-retryable errors. <br><br>See [Garbage collection and changefeeds]({% link {{ page.version.version }}/changefeed-messages.md %}#garbage-collection-and-changefeeds) for more detail on protecting changefeed data.<br><br>**Note:** If you use this option, changefeeds that are left paused for long periods of time can prevent garbage collection. Use with the [`gc_protect_expires_after`](#gc-protect-expire) option to set a limit for protected data and for how long a changefeed will remain paused.

src/current/v23.1/how-does-an-enterprise-changefeed-work.md

@@ -11,7 +11,7 @@ When an {{ site.data.products.enterprise }} changefeed is started on a node, tha
 {% include {{ page.version.version }}/cdc/work-distribution-setting.md %}
 {{site.data.alerts.end}}
 
-Each node uses its aggregator processors to send back checkpoint progress to the coordinator, which gathers this information to update the high-water mark timestamp. The high-water mark acts as a checkpoint for the changefeed’s job progress, and guarantees that all changes before (or at) the timestamp have been emitted. In the unlikely event that the changefeed’s coordinating node were to fail during the job, that role will move to a different node and the changefeed will restart from the last checkpoint. If restarted, the changefeed may [re-emit messages]({% link {{ page.version.version }}/changefeed-messages.md %}#duplicate-messages) starting at the high-water mark time to the current time. Refer to [Ordering Guarantees]({% link {{ page.version.version }}/changefeed-messages.md %}#ordering-and-delivery-guarantees) for detail on CockroachDB's at-least-once-delivery-guarantee and how per-key message ordering is applied.
+Each node uses its _aggregator processors_ to send back checkpoint progress to the coordinator, which gathers this information to update the _high-water mark timestamp_. The high-water mark acts as a checkpoint for the changefeed’s job progress, and guarantees that all changes before (or at) the timestamp have been emitted. In the unlikely event that the changefeed’s coordinating node were to fail during the job, that role will move to a different node and the changefeed will restart from the last checkpoint. If restarted, the changefeed may [re-emit messages]({% link {{ page.version.version }}/changefeed-messages.md %}#duplicate-messages) starting at the high-water mark time to the current time. Refer to [Ordering Guarantees]({% link {{ page.version.version }}/changefeed-messages.md %}#ordering-and-delivery-guarantees) for detail on CockroachDB's at-least-once-delivery-guarantee and how per-key message ordering is applied.
 
 <img src="{{ 'images/v23.1/changefeed-structure.png' | relative_url }}" alt="Changefeed process in a 3-node cluster" style="border:0px solid #eee;max-width:100%" />
 

src/current/v23.2/create-changefeed.md

@@ -176,7 +176,7 @@ Option | Value | Description
 <a name="lagging-ranges-threshold"></a><span class="version-tag">New in v23.2:</span> `lagging_ranges_threshold` | [Duration string](https://pkg.go.dev/time#ParseDuration) | Set 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`]({% link {{ page.version.version }}/monitor-and-debug-changefeeds.md %}#lagging-ranges-metric) metric. Note that ranges undergoing an [initial scan](#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.<br><br>**Default:** `3m`
 <a name="lagging-ranges-polling"></a><span class="version-tag">New in v23.2:</span> `lagging_ranges_polling_interval` | [Duration string](https://pkg.go.dev/time#ParseDuration) | Set 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.<br><br>**Default:** `1m`
 `metrics_label` | [`STRING`]({% link {{ page.version.version }}/string.md %}) | Define a metrics label to which the metrics for one or multiple changefeeds increment. All changefeeds also have their metrics aggregated.<br><br>The maximum length of a label is 128 bytes. There is a limit of 1024 unique labels.<br><br>`WITH metrics_label=label_name` <br><br>For more detail on usage and considerations, see [Using changefeed metrics labels]({% link {{ page.version.version }}/monitor-and-debug-changefeeds.md %}#using-changefeed-metrics-labels).
-<a name="min-checkpoint-frequency"></a>`min_checkpoint_frequency` | [Duration string](https://pkg.go.dev/time#ParseDuration) | Controls how often nodes flush their progress to the [coordinating changefeed node]({% link {{ page.version.version }}/how-does-an-enterprise-changefeed-work.md %}). Changefeeds will wait for at least the specified duration before a flush to the sink. This can help you control the flush frequency of higher latency sinks to achieve better throughput. However, more frequent checkpointing can increase CPU usage. If this is set to `0s`, a node will flush messages as long as the high-water mark has increased for the ranges that particular node is processing. If a changefeed is resumed, then `min_checkpoint_frequency` is the amount of time that changefeed will need to catch up. That is, it could emit [duplicate messages]({% link {{ page.version.version }}/changefeed-messages.md %}#duplicate-messages) during this time. <br><br>**Note:** [`resolved`](#resolved-option) messages will not be emitted more frequently than the configured `min_checkpoint_frequency` (but may be emitted less frequently). If you require `resolved` messages more frequently than `30s`, you must configure `min_checkpoint_frequency` to at least the desired `resolved` message frequency. For more details, refer to [Resolved message frequency]({% link {{ page.version.version }}/changefeed-messages.md %}#resolved-timestamp-frequency).<br><br>**Default:** `30s`
+<a name="min-checkpoint-frequency"></a>`min_checkpoint_frequency` | [Duration string](https://pkg.go.dev/time#ParseDuration) | Controls how often a node's changefeed [aggregator]({% link {{ page.version.version }}/how-does-an-enterprise-changefeed-work.md %}) will flush their progress to the [coordinating changefeed node]({% link {{ page.version.version }}/how-does-an-enterprise-changefeed-work.md %}). A node's changefeed aggregator will wait at least the specified duration between sending progress updates for the ranges it is watching to the coordinator. This can help you control the flush frequency of higher latency sinks to achieve better throughput. However, more frequent checkpointing can increase CPU usage. If this is set to `0s`, a node will flush messages as long as the high-water mark has increased for the ranges that particular node is processing. If a changefeed is resumed, then `min_checkpoint_frequency` is the amount of time that changefeed will need to catch up. That is, it could emit [duplicate messages]({% link {{ page.version.version }}/changefeed-messages.md %}#duplicate-messages) during this time. <br><br>**Note:** [`resolved`](#resolved-option) messages will not be emitted more frequently than the configured `min_checkpoint_frequency` (but may be emitted less frequently). If you require `resolved` messages more frequently than `30s`, you must configure `min_checkpoint_frequency` to at least the desired `resolved` message frequency. For more details, refer to [Resolved message frequency]({% link {{ page.version.version }}/changefeed-messages.md %}#resolved-timestamp-frequency).<br><br>**Default:** `30s`
 `mvcc_timestamp` | N/A |  Include the [MVCC]({% link {{ page.version.version }}/architecture/storage-layer.md %}#mvcc) timestamp for each emitted row in a changefeed. With the `mvcc_timestamp` option, each emitted row will always contain its MVCC timestamp, even during the changefeed's initial backfill.
 <a name="on-error"></a>`on_error` | `pause` / `fail` |  Use `on_error=pause` to pause the changefeed when encountering **non**-retryable errors. `on_error=pause` will pause the changefeed instead of sending it into a terminal failure state. **Note:** Retryable errors will continue to be retried with this option specified. <br><br>Use with [`protect_data_from_gc_on_pause`](#protect-pause) to protect changes from [garbage collection]({% link {{ page.version.version }}/configure-replication-zones.md %}#gc-ttlseconds).<br><br>If a changefeed with `on_error=pause` is running when a watched table is [truncated]({% link {{ page.version.version }}/truncate.md %}), the changefeed will pause but will not be able to resume reads from that table. Using [`ALTER CHANGEFEED`]({% link {{ page.version.version }}/alter-changefeed.md %}) to drop the table from the changefeed and then [resuming the job]({% link {{ page.version.version }}/resume-job.md %}) will work, but you cannot add the same table to the changefeed again. Instead, you will need to [create a new changefeed](#start-a-new-changefeed-where-another-ended) for that table.<br><br>Default: `on_error=fail`
 <a name="protect-pause"></a>`protect_data_from_gc_on_pause` | N/A |  This option is deprecated as of v23.2 and will be removed in a future release.<br><br>When a [changefeed is paused]({% link {{ page.version.version }}/pause-job.md %}), ensure that the data needed to [resume the changefeed]({% link {{ page.version.version }}/resume-job.md %}) is not garbage collected. If `protect_data_from_gc_on_pause` is **unset**, pausing the changefeed will release the existing protected timestamp records. It is also important to note that pausing and adding `protect_data_from_gc_on_pause` to a changefeed will not protect data if the [garbage collection]({% link {{ page.version.version }}/configure-replication-zones.md %}#gc-ttlseconds) window has already passed. <br><br>Use with [`on_error=pause`](#on-error) to protect changes from garbage collection when encountering non-retryable errors. <br><br>Refer to [Protect Changefeed Data from Garbage Collection]({% link {{ page.version.version }}/protect-changefeed-data.md %}) for more detail on protecting changefeed data.<br><br>**Note:** If you use this option, changefeeds that are left paused for long periods of time can prevent garbage collection. Use with the [`gc_protect_expires_after`](#gc-protect-expire) option to set a limit for protected data and for how long a changefeed will remain paused.