On this page Carat arrow pointing down
As of November 18, 2022, CockroachDB v21.1 is no longer supported. For more details, refer to the Release Support Policy.

CREATE CHANGEFEED is an Enterprise-only feature. For the core version, see EXPERIMENTAL CHANGEFEED FOR.

The CREATE CHANGEFEED statement creates a new Enterprise changefeed, which targets an allowlist of tables, called "watched rows". Every change to a watched row is emitted as a record in a configurable format (JSON or Avro) to a configurable sink (Kafka or a cloud storage sink). You can create, pause, resume, or cancel an Enterprise changefeed.

For more information, see Stream Data Out of CockroachDB Using Changefeeds.

Required privileges

To create a changefeed, the user must be a member of the admin role or have the CREATECHANGEFEED parameter set.


CREATE CHANGEFEED FOR TABLE table_name , INTO sink WITH option = value ,


Parameter Description
table_name The name of the table (or tables in a comma separated list) to create a changefeed for.

Note: Changefeeds do not share internal buffers, so each running changefeed will increase total memory usage. To watch multiple tables, we recommend creating a changefeed with a comma-separated list of tables.
sink The location of the configurable sink. The scheme of the URI indicates the type. For more information, see Sink URI below.
option / value For a list of available options and their values, see Options below.

Sink URI

The sink URI follows the basic format of:

URI Component Description
scheme The type of sink: kafka or any cloud storage sink.
host The sink's hostname or IP address.
port The sink's port.
query_parameters The sink's query parameters.


Example of a Kafka sink URI:

'kafka://{sasl user}&sasl_password={url-encoded password}&sasl_mechanism=SASL-SCRAM-SHA-256'

Cloud storage sink

Use a cloud storage sink to deliver changefeed data to OLAP or big data systems without requiring transport via Kafka.


Currently, cloud storage sinks only work with JSON and emits newline-delimited JSON files.

Example of a cloud storage sink URI:


Cloud storage sink URIs must be pre-pended with experimental- when working with changefeeds. For more information on the sink URI structure, see Use Cloud Storage for Bulk Operations.

Query parameters


Parameters should always be URI-encoded before they are included the changefeed's URI, as they often contain special characters. Use Javascript's encodeURIComponent function or Go language's url.QueryEscape function to URI-encode the parameters. Other languages provide similar functions to URI-encode special characters.

Query parameters include:

Sink Type
topic_prefix Kafka, cloud STRING Adds a prefix to all topic names.

For example, CREATE CHANGEFEED FOR TABLE foo INTO 'kafka://...?topic_prefix=bar_' would emit rows under the topic bar_foo instead of foo.
tls_enabled Kafka BOOL If true, enable Transport Layer Security (TLS) on the connection to Kafka. This can be used with a ca_cert (see below).

Default: false
ca_cert Kafka STRING The base64-encoded ca_cert file.

Note: To encode your ca.cert, run base64 -w 0 ca.cert.
client_cert Kafka STRING The base64-encoded Privacy Enhanced Mail (PEM) certificate. This is used with client_key.
client_key Kafka STRING The base64-encoded private key for the PEM certificate. This is used with client_cert.

Note: Client keys are often encrypted. You will receive an error if you pass an encrypted client key in your changefeed statement. To decrypt the client key, run: openssl rsa -in key.pem -out key.decrypt.pem -passin pass:{PASSWORD}. Once decrypted, be sure to update your changefeed statement to use the new key.decrypt.pem file instead.
sasl_enabled Kafka BOOL If true, the authentication protocol can be set to SCRAM or PLAIN using the sasl_mechanism parameter. You must have tls_enabled set to true to use SASL.

Default: false
sasl_mechanism Kafka STRING Can be set to SASL-SCRAM-SHA-256, SASL-SCRAM-SHA-512, or SASL-PLAIN. A sasl_user and sasl_password are required.

sasl_user Kafka STRING Your SASL username.
sasl_password Kafka STRING Your SASL password. Note: Passwords should be URL encoded since the value can contain characters that would cause authentication to fail.
file_size cloud STRING The file will be flushed (i.e., written to the sink) when it exceeds the specified file size. This can be used with the WITH resolved option, which flushes on a specified cadence.

Default: 16MB
insecure_tls_skip_verify Kafka BOOL If true, disable client-side validation of responses. Warning: Use this query parameter with caution, as it creates MITM vulnerabilities unless combined with another method of authentication.

Default: false


Option Value Description
updated N/A Include updated timestamps with each row.

If a cursor is provided, the "updated" timestamps will match the MVCC timestamps of the emitted rows, and there is no initial scan. If a cursor is not provided, the changefeed will perform an initial scan (as of the time the changefeed was created), and the "updated" timestamp for each change record emitted in the initial scan will be the timestamp of the initial scan. Similarly, when a backfill is performed for a schema change, the "updated" timestamp is set to the first timestamp for when the new schema is valid.
resolved Duration string Emits resolved timestamp events per changefeed in a format dependent on the connected sink. Resolved timestamp events do not emit until all ranges in the changefeed have progressed to a specific point in time.

Set an optional minimal duration between emitting resolved timestamps. Example: resolved='10s'. This option will only emit a resolved timestamp event if the timestamp has advanced and at least the optional duration has elapsed. If unspecified, all resolved timestamps are emitted as the high-water mark advances.
envelope key_only / wrapped Use key_only to emit only the key and no value, which is faster if you only want to know when the key changes.

Default: envelope=wrapped
cursor Timestamp Emit any changes after the given timestamp, but does not output the current state of the table first. If cursor is not specified, the changefeed starts by doing an initial scan of all the watched rows and emits the current value, then moves to emitting any changes that happen after the scan.

When starting a changefeed at a specific cursor, the cursor cannot be before the configured garbage collection window (see gc.ttlseconds) for the table you're trying to follow; otherwise, the changefeed will error. With default garbage collection settings, this means you cannot create a changefeed that starts more than 25 hours in the past.

cursor can be used to start a new changefeed where a previous changefeed ended.

Example: CURSOR='1536242855577149065.0000000000'
format json / experimental_avro Format of the emitted record. Currently, support for Avro is limited and experimental. For mappings of CockroachDB types to Avro types, see the table below.

Default: format=json.
confluent_schema_registry Schema Registry address The Schema Registry address is required to use experimental_avro.
key_in_value N/A Make the primary key 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 these sinks (currently only cloud storage sinks).
diff N/A Publish a before field with each message, which includes the value of the row before the update was applied.
compression gzip Compress changefeed data files written to a cloud storage sink. Currently, only Gzip is supported for compression.
protect_data_from_gc_on_pause N/A When a changefeed is paused, ensure that the data needed to resume the changefeed is not garbage collected.

Note: If you use this option, changefeeds left paused can prevent garbage collection for long periods of time.
schema_change_events default / column_changes The type of schema change event that triggers the behavior specified by the schema_change_policy option:
Default: schema_change_events=default
schema_change_policy backfill / nobackfill / stop The behavior to take when an event specified by the schema_change_events option occurs:
  • backfill: When schema changes with column backfill are finished, output all watched rows using the new schema.
  • nobackfill: For schema changes with column backfill, perform no logical backfills. The changefeed will still emit any duplicate records for the table being altered, but will not emit the new schema records.
  • stop: schema changes with column backfill, wait for all data preceding the schema change to be resolved before exiting with an error indicating the timestamp at which the schema change occurred. An error: schema change occurred at <timestamp> will display in the cockroach.log file.

Default: schema_change_policy=backfill
initial_scan / no_initial_scan N/A Control whether or not an initial scan will occur at the start time of a changefeed. initial_scan and no_initial_scan cannot be used simultaneously. If neither initial_scan nor no_initial_scan is specified, an initial scan will occur if there is no cursor, and will not occur if there is one. This preserves the behavior from previous releases.

Default: initial_scan
If used in conjunction with cursor, an initial scan will be performed at the cursor timestamp. If no cursor is specified, the initial scan is performed at now().
full_table_name N/A New in v21.1: Use fully-qualified table name in topics, subjects, schemas, and record output instead of the default table name. This can prevent unintended behavior when the same table name is present in multiple databases.

Example: CREATE CHANGEFEED FOR foo... WITH full_table_name will create the topic name instead of foo.
avro_schema_prefix Schema prefix name New in v21.1: Provide a namespace for the schema of a table in addition to the default, the table name. This allows multiple databases or clusters to share the same schema registry when the same table name is present in multiple databases.

Example: CREATE CHANGEFEED FOR foo WITH format=avro, confluent_schema_registry='registry_url', avro_schema_prefix='super' will register subjects as superfoo-key and superfoo-value with the namespace super.
kafka_sink_config STRING Set fields to configure the required level of message acknowledgement from the Kafka server, the version of the server, and batching parameters for Kafka sinks. See Advanced Configuration for more detail on configuring all the available fields for this option.

Example: CREATE CHANGEFEED FOR table INTO 'kafka://localhost:9092' WITH kafka_sink_config='{"Flush": {"MaxMessages": 1, "Frequency": "1s"}, "RequiredAcks": "ONE"}'

Using the format=experimental_avro, envelope=key_only, and updated options together is rejected. envelope=key_only prevents any rows with updated fields from being emitted, which makes the updated option meaningless.

Avro limitations

Currently, support for Avro is limited and experimental.

Below are clarifications for particular SQL types and values for Avro changefeeds:

  • Decimals must have precision specified.
  • BIT and VARBIT types are encoded as arrays of 64-bit integers.

    For efficiency, CockroachDB encodes BIT and VARBIT bitfield types as arrays of 64-bit integers. That is, base-2 (binary format) BIT and VARBIT data types are converted to base 10 and stored in arrays. Encoding in CockroachDB is big-endian, therefore the last value may have many trailing zeroes. For this reason, the first value of each array is the number of bits that are used in the last value of the array.

    For instance, if the bitfield is 129 bits long, there will be 4 integers in the array. The first integer will be 1; representing the number of bits in the last value, the second integer will be the first 64 bits, the third integer will be bits 65–128, and the last integer will either be 0 or 9223372036854775808 (i.e. the integer with only the first bit set, or 1000000000000000000000000000000000000000000000000000000000000000 when base 2).

    This example is base-10 encoded into an array as follows:

    {"array": [1, <first 64 bits>, <second 64 bits>, 0 or 9223372036854775808]}

    For downstream processing, it is necessary to base-2 encode every element in the array (except for the first element). The first number in the array gives you the number of bits to take from the last base-2 number — that is, the most significant bits. So, in the example above this would be 1. Finally, all the base-2 numbers can be appended together, which will result in the original number of bits, 129.

    In a different example of this process where the bitfield is 136 bits long, the array would be similar to the following when base-10 encoded:

    {"array": [8, 18293058736425533439, 18446744073709551615, 13690942867206307840]}

    To then work with this data, you would convert each of the elements in the array to base-2 numbers, besides the first element. For the above array, this would convert to:

    [8, 1111110111011011111111111111111111111111111111111111111111111111, 1111111111111111111111111111111111111111111111111111111111111111, 1011111000000000000000000000000000000000000000000000000000000000]

    Next, you use the first element in the array to take the number of bits from the last base-2 element, 10111110. Finally, you append each of the base-2 numbers together — in the above array, the second, third, and truncated last element. This results in 136 bits, the original number of bits.

Avro types

Below is a mapping of CockroachDB types to Avro types:

CockroachDB Type Avro Type Avro Logical Type
BIT Array of LONG

Topic Naming

By default, a Kafka topic has the same name as the table that a changefeed was created on. If a changefeed was created on multiple tables, the changefeed will write to multiple topics corresponding to those table names. You can specify a topic prefix or use the full_table_name option to modify this.

You can either manually create a topic in your Kafka cluster before starting the changefeed, or the topic will be automatically created when the changefeed connects to your Kafka cluster.


You must have the Kafka cluster setting auto.create.topics.enable set to true for automatic topic creation. This will create the topic when the changefeed sends its first message. If you create the consumer before that, you will also need the Kafka consumer configuration to be set to true.

Kafka has the following topic limitations:

  • Legal characters are numbers, letters, and [._-].
  • The maximum character length of a topic name is 249.
  • Topics with a period (.) and underscore (_) can collide on internal Kafka data structures, so you should use either but not both.
  • Characters not accepted by Kafka will be automatically encoded as unicode characters by CockroachDB.

Advanced changefeed configuration


The configurations and settings explained in these sections will have significant impact on a changefeed's behavior.

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

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.
  • Determine what a successful write to Kafka is with the kafka_sink_config: {"RequiredAcks": "ALL"} option, which provides the highest consistency level.
  • Set the changefeed.memory.per_changefeed_limit cluster setting to a higher limit to give more memory for buffering per single changefeed. Increasing this limit is useful to avoid errors in situations where the sink is unreachable and the memory buffer fills up.
  • Use Kafka or cloud storage sinks when tuning for high durability in changefeeds.
  • 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:

  • Set the resolved option to a higher duration (e.g. 10 minutes or more) 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 to avoid an error due to the memory buffer overfilling.
  • 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.

Kafka Sink Configuration

The kafka_sink_config option allows configuration of a changefeed's message delivery, Kafka server version, and batching parameters.


Each of the following settings have significant impact on a changefeed's behavior, such as latency. For example, it is possible to configure batching parameters to be very high, which would negatively impact changefeed latency. As a result it would take a long time to see messages coming through to the sink. Also, large batches may be rejected by the Kafka server unless it's separately configured to accept a high max.message.bytes.

kafka_sink_config='{"Flush": {"MaxMessages": 1, "Frequency": "1s"}, "Version": "", "RequiredAcks": "ONE" }'

The configurable fields include:

  • "Flush"."MaxMessages" and "Flush"."Frequency": These are configurable batching parameters depending on latency and throughput needs. For example, if "MaxMessages" is set to 1000 and "Frequency" to 1 second, it will flush to Kafka either after 1 second or after 1000 messages are batched, whichever comes first. It's important to consider that if there are not many messages then a "1s" frequency will add 1 second latency. However, if there is a larger influx of messages these will be flushed quicker.

    • "MaxMessages": sets the maximum number of messages the producer will send in a single broker request. Increasing this value allows all messages to be sent in a batch. Default: 1. Type: INT.
    • "Frequency": sets the maximum time that messages will be batched. Default: "0s". Type: INTERVAL.
  • "Version": sets the appropriate Kafka cluster version, which can be used to connect to Kafka versions < v1.0 (kafka_sink_config='{"Version": ""}'). Default: "" Type: STRING.

  • "RequiredAcks": specifies what a successful write to Kafka is. CockroachDB guarantees at least once delivery of messages — this value defines the delivery. Type: STRING. The possible values are:
    • "ONE": a write to Kafka is successful once the leader node has committed and acknowledged the write. Note that this has the potential risk of dropped messages; if the leader node acknowledges before replicating to a quorum of other Kafka nodes, but then fails. This is the default value.
    • "NONE": no Kafka brokers are required to acknowledge that they have committed the message. This will decrease latency and increase throughput, but comes at the cost of lower consistency.
    • "ALL": a quorum must be reached (that is, most Kafka brokers have committed the message) before the leader can acknowledge. This is the highest consistency level.



The messages (i.e., keys and values) emitted to a sink are specific to the envelope. The default format is wrapped, and the output messages are composed of the following:

  • Key: An array always composed of the row's PRIMARY KEY field(s) (e.g., [1] for JSON or {"id":{"long":1}} for Avro).
  • Value:
    • One of three possible top-level fields:
      • after, which contains the state of the row after the update (or null' for DELETEs).
      • updated, which contains the updated timestamp.
      • resolved, which is emitted for records representing resolved timestamps. These records do not include an "after" value since they only function as checkpoints.
    • For INSERT and UPDATE, the current state of the row inserted or updated.
    • For DELETE, null.

For example:

Statement Response
INSERT INTO office_dogs VALUES (1, 'Petee'); JSON: [1] {"after": {"id": 1, "name": "Petee"}}
Avro: {"id":{"long":1}} {"after":{"office_dogs":{"id":{"long":1},"name":{"string":"Petee"}}}}
DELETE FROM office_dogs WHERE name = 'Petee' JSON: [1] {"after": null}
Avro: {"id":{"long":1}} {"after":null}


The files emitted to a sink use the following naming conventions:



General file format


For example:


Resolved file format


For example:



Create a changefeed connected to Kafka

> CREATE CHANGEFEED FOR TABLE name, name2, name3
  INTO 'kafka://host:port'
  WITH updated, resolved;
|       job_id       |
| 360645287206223873 |
(1 row)

For more information on how to create a changefeed connected to Kafka, see Stream Data Out of CockroachDB Using Changefeeds.

Create a changefeed connected to Kafka using Avro

> CREATE CHANGEFEED FOR TABLE name, name2, name3
  INTO 'kafka://host:port'
  WITH format = experimental_avro, confluent_schema_registry = <schema_registry_address>;
|       job_id       |
| 360645287206223873 |
(1 row)

For more information on how to create a changefeed that emits an Avro record, see Stream Data Out of CockroachDB Using Changefeeds.

Create a changefeed connected to a cloud storage sink


This is an experimental feature. The interface and output are subject to change.

> CREATE CHANGEFEED FOR TABLE name, name2, name3
  INTO 'experimental-scheme://host?parameters'
  WITH updated, resolved;
|       job_id       |
| 360645287206223873 |
(1 row)

For more information on how to create a changefeed connected to a cloud storage sink, see Stream Data Out of CockroachDB Using Changefeeds.

Manage a changefeed

Use the following SQL statements to pause, resume, and cancel a changefeed.


Changefeed-specific SQL statements (e.g., CANCEL CHANGEFEED) will be added in the future.

Pause a changefeed

> PAUSE JOB job_id;

For more information, see PAUSE JOB.

Resume a paused changefeed

> RESUME JOB job_id;

For more information, see RESUME JOB.

Cancel a changefeed

> CANCEL JOB job_id;

For more information, see CANCEL JOB.

Configuring all changefeeds

It is useful to be able to pause all running changefeeds during troubleshooting, testing, or when a decrease in CPU load is needed.

To pause all running changefeeds:

PAUSE JOBS (SELECT job_id FROM [SHOW JOBS] WHERE job_type='CHANGEFEED' AND status IN ('running'));

This will change the status for each of the running changefeeds to paused, which can be verified with SHOW JOBS.

To resume all running changefeeds:

RESUME JOBS (SELECT job_id FROM [SHOW JOBS] WHERE job_type='CHANGEFEED' AND status IN ('paused'));

This will resume the changefeeds and update the status for each of the changefeeds to running.

Start a new changefeed where another ended

Find the high-water timestamp for the ended changefeed:

> SELECT * FROM WHERE job_id = <job_id>;
        job_id       |  job_type  | ... |      high_water_timestamp      | error | coordinator_id
+--------------------+------------+ ... +--------------------------------+-------+----------------+
  383870400694353921 | CHANGEFEED | ... | 1537279405671006870.0000000000 |       |              1
(1 row)

Use the high_water_timestamp to start the new changefeed:

> CREATE CHANGEFEED FOR TABLE name, name2, name3
  INTO 'kafka//host:port'
  WITH cursor = '<high_water_timestamp>';

Note that because the cursor is provided, the initial scan is not performed.

See also

Yes No
On this page

Yes No