Skip to main content

redpanda

A Kafka input using the Franz Kafka client library.

# Common config fields, showing default values
input:
  label: ""
  redpanda:
    seed_brokers: [] # No default (optional)
    topics: [] # No default (optional)
    regexp_topics_include: [] # No default (optional)
    regexp_topics_exclude: [] # No default (optional)
    transaction_isolation_level: "read_uncommitted"
    consumer_group: "" # No default (optional)
    auto_replay_nacks: true

When a consumer group is specified this input consumes one or more topics where partitions will automatically balance across any other connected clients with the same consumer group. When a consumer group is not specified topics can either be consumed in their entirety or with explicit partitions.

Delivery Guarantees​

When using consumer groups the offsets of "delivered" records will be committed automatically and continuously, and in the event of restarts these committed offsets will be used in order to resume from where the input left off. Expanso Edge guarantees at least once delivery by ensuring that records are only considered to be delivered when all configured outputs that the record is routed to have confirmed delivery.

Ordering​

In order to preserve ordering of topic partitions, records consumed from each partition are processed and delivered in the order that they are received, and only one batch of records of a given partition will ever be processed at a time. This means that parallel processing can only occur when multiple topic partitions are being consumed, but ensures that data is processed in a sequential order as determined from the source partition.

However, one way in which the order of records can be mixed is when delivery errors occur and error handling mechanisms kick in. Expanso Edge always leans towards at least once delivery unless instructed otherwise, and this includes reattempting delivery of data when the ordering of that data can no longer be guaranteed.

For example, a batch of records may have been sent to an output broker and only a subset of records were delivered, in this case Expanso Edge by default will reattempt to deliver the records that failed, even though these failed records may have come before records that were previously delivered successfully.

In order to avoid this scenario you must specify in your configuration an alternative way to handle delivery errors in the form of a fallback output. It is good practice to also disable the field auto_retry_nacks by setting it to false when you've added an explicit fallback output as this will improve the throughput of your pipeline. For example, the following config avoids ordering issues by specifying a fallback output into a DLQ topic, which is also retried indefinitely as a way to apply back pressure during connectivity issues:

output:
  fallback:
    - redpanda:
        seed_brokers: [ localhost:9092 ]
        topic: foo
    - retry:
        output:
          redpanda:
            seed_brokers: [ localhost:9092 ]
            topic: foo_dlq

Batching​

Records are processed and delivered from each partition in batches as received from brokers. These batch sizes are therefore dynamically sized in order to optimise throughput, but can be tuned with the config field max_yield_batch_bytes, or unordered_processing.batching when unordered processing is enabled. Batches can be further broken down using the split processor.

Metrics​

Emits a redpanda_lag metric with topic and partition labels for each consumed topic.

Metadata​

This input adds the following metadata fields to each message:

- kafka_key
- kafka_topic
- kafka_partition
- kafka_offset
- kafka_lag
- kafka_timestamp_ms
- kafka_timestamp_unix
- kafka_tombstone_message
- All record headers

Fields​

seed_brokers​

A list of broker addresses to connect to in order to establish connections. If an item of the list contains commas it will be expanded into multiple addresses. When this field is omitted the global redpanda block will be referenced for connection details.

Type: array of string

client_id​

An identifier for the client connection.

Type: string
Default: "redpanda-connect"

tls​

Custom TLS settings can be used to override system defaults.

Type: object

tls.enabled​

Whether custom TLS settings are enabled.

Type: bool
Default: false

tls.skip_cert_verify​

Whether to skip server side certificate verification.

Type: bool
Default: false

tls.enable_renegotiation​

Whether to allow the remote server to repeatedly request renegotiation. Enable this option if you're seeing the error message local error: tls: no renegotiation.

Type: bool
Default: false

tls.root_cas​

An optional root certificate authority to use. This is a string, representing a certificate chain from the parent trusted root certificate, to possible intermediate signing certificates, to the host certificate.

Secret

This field contains sensitive information. Use a secret reference rather than a literal value.

Type: string
Default: ""

tls.root_cas_file​

An optional path of a root certificate authority file to use. This is a file, often with a .pem extension, containing a certificate chain from the parent trusted root certificate, to possible intermediate signing certificates, to the host certificate.

Type: string
Default: ""

tls.client_certs​

A list of client certificates to use. For each certificate either the fields cert and key, or cert_file and key_file should be specified, but not both.

Type: array of object
Default: []

tls.client_certs[].cert​

A plain text certificate to use.

Type: string
Default: ""

tls.client_certs[].key​

A plain text certificate key to use.

Secret

This field contains sensitive information. Use a secret reference rather than a literal value.

Type: string
Default: ""

tls.client_certs[].cert_file​

The path of a certificate to use.

Type: string
Default: ""

tls.client_certs[].key_file​

The path of a certificate key to use.

Type: string
Default: ""

tls.client_certs[].password​

A plain text password for when the private key is password encrypted in PKCS#1 or PKCS#8 format. The obsolete pbeWithMD5AndDES-CBC algorithm is not supported for the PKCS#8 format.

Because the obsolete pbeWithMD5AndDES-CBC algorithm does not authenticate the ciphertext, it is vulnerable to padding oracle attacks that can let an attacker recover the plaintext.

Secret

This field contains sensitive information. Use a secret reference rather than a literal value.

Type: string
Default: ""

sasl​

Specify one or more methods of SASL authentication. SASL is tried in order; if the broker supports the first mechanism, all connections will use that mechanism. If the first mechanism fails, the client will pick the first supported mechanism. If the broker does not support any client mechanisms, connections will fail.

Type: array of object

sasl[].mechanism​

The SASL mechanism to use.

Type: string

OptionSummary
AWS_MSK_IAMAWS IAM based authentication as specified by the 'aws-msk-iam-auth' java library.
OAUTHBEAREROAuth Bearer based authentication.
PLAINPlain text authentication.
REDPANDA_CLOUD_SERVICE_ACCOUNTRedpanda Cloud Service Account authentication when running in Redpanda Cloud.
SCRAM-SHA-256SCRAM based authentication as specified in RFC5802.
SCRAM-SHA-512SCRAM based authentication as specified in RFC5802.
noneDisable sasl authentication

sasl[].username​

A username to provide for PLAIN or SCRAM-* authentication.

Type: string
Default: ""

sasl[].password​

A password to provide for PLAIN or SCRAM-* authentication.

Secret

This field contains sensitive information. Use a secret reference rather than a literal value.

Type: string
Default: ""

sasl[].token​

The token to use for a single session's OAUTHBEARER authentication.

Secret

This field contains sensitive information. Use a secret reference rather than a literal value.

Type: string
Default: ""

sasl[].extensions​

Key/value pairs to add to OAUTHBEARER authentication requests.

Type: map of string

sasl[].aws​

Contains AWS specific fields for when the mechanism is set to AWS_MSK_IAM.

Type: object

sasl[].aws.region​

The AWS region to target.

Type: string

sasl[].aws.endpoint​

Allows you to specify a custom endpoint for the AWS API.

Type: string

sasl[].aws.tcp​

TCP socket configuration.

Type: object

sasl[].aws.tcp.connect_timeout​

Maximum amount of time a dial will wait for a connect to complete. Zero disables.

Type: string
Default: "0s"

sasl[].aws.tcp.keep_alive​

TCP keep-alive probe configuration.

Type: object

sasl[].aws.tcp.keep_alive.idle​

Duration the connection must be idle before sending the first keep-alive probe. Zero defaults to 15s. Negative values disable keep-alive probes.

Type: string
Default: "15s"

sasl[].aws.tcp.keep_alive.interval​

Duration between keep-alive probes. Zero defaults to 15s.

Type: string
Default: "15s"

sasl[].aws.tcp.keep_alive.count​

Maximum unanswered keep-alive probes before dropping the connection. Zero defaults to 9.

Type: int
Default: 9

sasl[].aws.tcp.tcp_user_timeout​

Maximum time to wait for acknowledgment of transmitted data before killing the connection. Linux-only (kernel 2.6.37+), ignored on other platforms. When enabled, keep_alive.idle must be greater than this value per RFC 5482. Zero disables.

Type: string
Default: "0s"

sasl[].aws.credentials​

Optional manual configuration of AWS credentials to use. More information can be found in this document.

Type: object

sasl[].aws.credentials.profile​

A profile from ~/.aws/credentials to use.

Type: string

sasl[].aws.credentials.id​

The ID of credentials to use.

Type: string

sasl[].aws.credentials.secret​

The secret for the credentials being used.

Secret

This field contains sensitive information. Use a secret reference rather than a literal value.

Type: string

sasl[].aws.credentials.token​

The token for the credentials being used, required when using short term credentials.

Secret

This field contains sensitive information. Use a secret reference rather than a literal value.

Type: string

sasl[].aws.credentials.from_ec2_role​

Use the credentials of a host EC2 machine configured to assume an IAM role associated with the instance.

Type: bool

sasl[].aws.credentials.role​

A role ARN to assume.

Type: string

sasl[].aws.credentials.role_external_id​

An external ID to provide when assuming a role.

Type: string

metadata_max_age​

The maximum age of metadata before it is refreshed. This interval also controls how frequently regex topic patterns are re-evaluated to discover new matching topics.

Type: string
Default: "1m"

request_timeout_overhead​

The request time overhead. Uses the given time as overhead while deadlining requests. Roughly equivalent to request.timeout.ms, but grants additional time to requests that have timeout fields.

Type: string
Default: "10s"

conn_idle_timeout​

The rough amount of time to allow connections to idle before they are closed.

Type: string
Default: "20s"

tcp​

TCP socket configuration.

Type: object

tcp.connect_timeout​

Maximum amount of time a dial will wait for a connect to complete. Zero disables.

Type: string
Default: "0s"

tcp.keep_alive​

TCP keep-alive probe configuration.

Type: object

tcp.keep_alive.idle​

Duration the connection must be idle before sending the first keep-alive probe. Zero defaults to 15s. Negative values disable keep-alive probes.

Type: string
Default: "15s"

tcp.keep_alive.interval​

Duration between keep-alive probes. Zero defaults to 15s.

Type: string
Default: "15s"

tcp.keep_alive.count​

Maximum unanswered keep-alive probes before dropping the connection. Zero defaults to 9.

Type: int
Default: 9

tcp.tcp_user_timeout​

Maximum time to wait for acknowledgment of transmitted data before killing the connection. Linux-only (kernel 2.6.37+), ignored on other platforms. When enabled, keep_alive.idle must be greater than this value per RFC 5482. Zero disables.

Type: string
Default: "0s"

topics​

A list of topics to consume from. Multiple comma separated topics can be listed in a single element. When a consumer_group is specified partitions are automatically distributed across consumers of a topic, otherwise all partitions are consumed.

Alternatively, it's possible to specify explicit partitions to consume from with a colon after the topic name, e.g. foo:0 would consume the partition 0 of the topic foo. This syntax supports ranges, e.g. foo:0-10 would consume partitions 0 through to 10 inclusive.

Finally, it's also possible to specify an explicit offset to consume from by adding another colon after the partition, e.g. foo:0:10 would consume the partition 0 of the topic foo starting from the offset 10. If the offset is not present (or remains unspecified) then the field start_from_oldest determines which offset to start from.

Type: array of string

regexp_topics_include​

A list of regular expression patterns for matching topics to consume from. When specified, the client will periodically refresh the list of matching topics based on the metadata_max_age interval. This enables regex mode and cannot be used together with the topics field. Use regexp_topics_exclude to exclude specific patterns.

Type: array of string

regexp_topics_exclude​

A list of regular expression patterns for excluding topics when regex mode is enabled (via regexp_topics or regexp_topics_include). Topics matching any of these patterns will be excluded from consumption, even if they match include patterns.

Type: array of string

rack_id​

A rack specifies where the client is physically located and changes fetch requests to consume from the closest replica as opposed to the leader replica.

Type: string
Default: ""

instance_id​

When using a consumer group, an instance ID specifies the groups static membership, which can prevent rebalances during reconnects. When using a instance ID the client does NOT leave the group when closing. To actually leave the group one must use an external admin command to leave the group on behalf of this instance ID. This ID must be unique per consumer within the group.

Type: string
Default: ""

rebalance_timeout​

When using a consumer group, rebalance_timeout sets how long group members are allowed to take when a rebalance has begun. This timeout is how long all members are allowed to complete work and commit offsets, minus the time it took to detect the rebalance (from a heartbeat).

Type: string
Default: "45s"

session_timeout​

When using a consumer group, session_timeout sets how long a member in the group can go between heartbeats. If a member does not heartbeat in this timeout, the broker will remove the member from the group and initiate a rebalance.

Type: string
Default: "1m"

heartbeat_interval​

When using a consumer group, heartbeat_interval sets how long a group member goes between heartbeats to Kafka. Kafka uses heartbeats to ensure that a group member's session stays active. This value should be no higher than 1/3rd of the session_timeout. This is equivalent to the Java heartbeat.interval.ms setting.

Type: string
Default: "3s"

start_offset​

Sets the offset to start consuming from, or if OffsetOutOfRange is seen while fetching, to restart consuming from.

Type: string
Default: "earliest"

OptionSummary
committedPrevents consuming a partition in a group if the partition has no prior commits. Corresponds to Kafka's auto.offset.reset=none option
earliestStart from the earliest offset. Corresponds to Kafka's auto.offset.reset=earliest option.
latestStart from the latest offset. Corresponds to Kafka's auto.offset.reset=latest option.

fetch_max_bytes​

Sets the maximum amount of bytes a broker will try to send during a fetch. Note that brokers may not obey this limit if it has records larger than this limit. This is the equivalent to the Java fetch.max.bytes setting.

Type: string
Default: "50MiB"

fetch_max_wait​

Sets the maximum amount of time a broker will wait for a fetch response to hit the minimum number of required bytes. This is the equivalent to the Java fetch.max.wait.ms setting.

Type: string
Default: "5s"

fetch_min_bytes​

Sets the minimum amount of bytes a broker will try to send during a fetch. This is the equivalent to the Java fetch.min.bytes setting.

Type: string
Default: "1B"

fetch_max_partition_bytes​

Sets the maximum amount of bytes that will be consumed for a single partition in a fetch request. Note that if a single batch is larger than this number, that batch will still be returned so the client can make progress. This is the equivalent to the Java fetch.max.partition.bytes setting.

Type: string
Default: "1MiB"

transaction_isolation_level​

The transaction isolation level

Type: string
Default: "read_uncommitted"

OptionSummary
read_committedIf set, only committed transactional records are processed.
read_uncommittedIf set, then uncommitted records are processed.

consumer_group​

An optional consumer group to consume as. When specified the partitions of specified topics are automatically distributed across consumers sharing a consumer group, and partition offsets are automatically committed and resumed under this name. Consumer groups are not supported when specifying explicit partitions to consume from in the topics field.

Type: string

commit_period​

The period of time between each commit of the current partition offsets. Offsets are always committed during shutdown.

Type: string
Default: "5s"

partition_buffer_bytes​

A buffer size (in bytes) for each consumed partition, allowing records to be queued internally before flushing. Increasing this may improve throughput at the cost of higher memory utilisation. Note that each buffer can grow slightly beyond this value.

Type: string
Default: "1MB"

topic_lag_refresh_period​

The period of time between each topic lag refresh cycle.

Type: string
Default: "5s"

max_yield_batch_bytes​

The maximum size (in bytes) for each batch yielded by this input. This value must be less than or equal to the partition_buffer_bytes. If using Redpanda output, this value should not be greater than the max_message_bytes option value (1MB by default), and for high-throughput scenarios they should be equal.

Type: string
Default: "32KB"

unordered_processing​

Configures partition consumers to allow parallel and therefore unordered processing of messages of any given partition. This allows for better utilization of processing threads and asynchronous publishing at the output level. The maximum parallelization of each partition is determined by the checkpoint_limit field.

Type: object

unordered_processing.enabled​

Whether to enable the unordered processing of messages from a given partition.

Type: bool
Default: false

unordered_processing.checkpoint_limit​

Determines how many messages of the same partition can be processed in parallel before applying back pressure. When a message of a given offset is delivered to the output the offset is only allowed to be committed when all messages of prior offsets have also been delivered, this ensures at-least-once delivery guarantees. However, this mechanism also increases the likelihood of duplicates in the event of crashes or server faults, reducing the checkpoint limit will mitigate this.

Type: int
Default: 1024

unordered_processing.batching​

Allows you to configure a batching policy that applies to individual topic partitions in order to batch messages together before flushing them for processing. Batching can be beneficial for performance as well as useful for windowed processing, and doing so this way preserves the ordering of topic partitions.

Type: object

unordered_processing.batching.count​

A number of messages at which the batch should be flushed. If 0 disables count based batching.

Type: int
Default: 0

unordered_processing.batching.byte_size​

An amount of bytes at which the batch should be flushed. If 0 disables size based batching.

Type: int
Default: 0

unordered_processing.batching.period​

A period in which an incomplete batch should be flushed regardless of its size.

Type: string
Default: ""

unordered_processing.batching.check​

A Bloblang query that should return a boolean value indicating whether a message should end a batch.

Type: string
Default: ""

unordered_processing.batching.processors​

A list of processors to apply to a batch as it is flushed. This allows you to aggregate and archive the batch however you see fit. Please note that all resulting messages are flushed as a single batch, therefore splitting the batch into smaller batches using these processors is a no-op.

Type: array of processor

auto_replay_nacks​

Whether messages that are rejected (nacked) at the output level should be automatically replayed indefinitely, eventually resulting in back pressure if the cause of the rejections is persistent. If set to false these messages will instead be deleted. Disabling auto replays can greatly improve memory efficiency of high throughput streams as the original shape of the data can be discarded immediately upon consumption and mutation.

Type: bool
Default: true

timely_nacks_maximum_wait​

EXPERIMENTAL: Specify a maximum period of time in which each message can be consumed and awaiting either acknowledgement or rejection before rejection is instead forced. This can be useful for avoiding situations where certain downstream components can result in blocked confirmation of delivery that exceeds SLAs.

Type: string

extract_tracing_map​

EXPERIMENTAL: A Bloblang mapping that attempts to extract an object containing tracing propagation information, which will then be used as the root tracing span for the message. The specification of the extracted fields must match the format used by the service wide tracer.

Type: string

Superseded offset and topic-pattern fields​

Field descriptions above mention start_from_oldest and regexp_topics, which are deprecated and therefore not listed among the fields. They still work, so an existing config using them keeps validating and deploying, but new configs should use the replacements:

DeprecatedUse instead
start_from_oldest: true (boolean)start_offset: earliest
start_from_oldest: falsestart_offset: latest
regexp_topics: true (boolean, treating topics entries as patterns)regexp_topics_include (a list of patterns, with regexp_topics_exclude to subtract from it)

So where the topics description says start_from_oldest decides the starting offset, start_offset is the field to set; and "regex mode" is enabled by populating regexp_topics_include. start_offset also takes committed, which the boolean could not express: it refuses to consume a partition that has no prior commits in the group, rather than falling back to either end of the log.

One name above is not merely deprecated but gone: the ordering guidance recommends disabling auto_retry_nacks when you add a fallback output. That field is now auto_replay_nacks, and the old name is rejected at submission with Unknown field 'auto_retry_nacks', so copy the advice with the current name:

input:
  redpanda:
    seed_brokers: [localhost:9092]
    topics: [events]
    auto_replay_nacks: false   # pair with an explicit fallback output