aws_kinesis_streams sink

Batches `log` events to AWS Kinesis Data Stream via the `PutRecords` API endpoint.

The aws_kinesis_streams sink is in beta. Please see the current enhancements and bugs for known issues. We kindly ask that you add any missing issues as it will help shape the roadmap of this component.

The aws_kinesis_streams sink batches log events to AWS Kinesis Data Stream via the PutRecords API endpoint.

Config File

vector.toml (example)
vector.toml (schema)
vector.toml (specification)
[sinks.my_sink_id]
# REQUIRED - General
type = "aws_kinesis_streams" # must be: "aws_kinesis_streams"
inputs = ["my-source-id"]
region = "us-east-1"
stream_name = "my-stream"
# OPTIONAL - General
healthcheck = true # default
hostname = "127.0.0.0:5000"
partition_key_field = "user_id" # no default
# OPTIONAL - Batching
batch_size = 1049000 # default, bytes
batch_timeout = 1 # default, seconds
# OPTIONAL - Requests
encoding = "json" # no default, enum: "json" or "text"
rate_limit_duration = 1 # default, seconds
rate_limit_num = 5 # default
request_in_flight_limit = 5 # default
request_timeout_secs = 30 # default, seconds
retry_attempts = 5 # default
retry_backoff_secs = 5 # default, seconds
# OPTIONAL - Buffer
[sinks.my_sink_id.buffer]
type = "memory" # default, enum: "memory" or "disk"
when_full = "block" # default, enum: "block" or "drop_newest"
max_size = 104900000 # no default, bytes, relevant when type = "disk"
num_items = 500 # default, events, relevant when type = "memory"

Options

Key

Type

Description

REQUIRED - General

type

string

The component type required must be: "aws_kinesis_streams"

inputs

[string]

A list of upstream source or transform IDs. See Config Composition for more info. required example: ["my-source-id"]

region

string

The AWS region of the target Kinesis stream resides. required example: "us-east-1"

stream_name

string

The stream name of the target Kinesis Logs stream. required example: "my-stream"

OPTIONAL - General

healthcheck

bool

Enables/disables the sink healthcheck upon start. See Health Checks for more info. default: true

hostname

string

Custom hostname to send requests to. Useful for testing. default: "<aws-service-hostname>"

partition_key_field

string

The event field used as the Kinesis record's partition key value. See Partitioning for more info. no default example: "user_id"

OPTIONAL - Batching

batch_size

int

The maximum size of a batch before it is flushed. See Buffers & Batches for more info. default: 1049000 unit: bytes

batch_timeout

int

The maximum age of a batch before it is flushed. See Buffers & Batches for more info. default: 1 unit: seconds

OPTIONAL - Requests

encoding

string

The encoding format used to serialize the events before flushing. The default is dynamic based on if the event is structured or not. See Encodings for more info. no default enum: "json" or "text"

rate_limit_duration

int

The window used for the request_rate_limit_num option See Rate Limits for more info. default: 1 unit: seconds

rate_limit_num

int

The maximum number of requests allowed within the rate_limit_duration window. See Rate Limits for more info. default: 5

request_in_flight_limit

int

The maximum number of in-flight requests allowed at any given time. See Rate Limits for more info. default: 5

request_timeout_secs

int

The maximum time a request can take before being aborted. See Timeouts for more info. default: 30 unit: seconds

retry_attempts

int

The maximum number of retries to make for failed requests. See Retry Policy for more info. default: 5

retry_backoff_secs

int

The amount of time to wait before attempting a failed request again. See Retry Policy for more info. default: 5 unit: seconds

OPTIONAL - Buffer

buffer.type

string

The buffer's type / location. disk buffers are persistent and will be retained between restarts. default: "memory" enum: "memory" or "disk"

buffer.when_full

string

The behavior when the buffer becomes full. default: "block" enum: "block" or "drop_newest"

buffer.max_size

int

The maximum size of the buffer on the disk. Only relevant when type = "disk" no default example: 104900000 unit: bytes

buffer.num_items

int

The maximum number of events allowed in the buffer. Only relevant when type = "memory" default: 500 unit: events

Examples

The aws_kinesis_streams sink batches log up to the batch_size or batch_timeout options. When flushed, Vector will write to AWS Kinesis Data Stream via the PutRecords API endpoint. The encoding is dictated by the encoding option. For example:

POST / HTTP/1.1
Host: kinesis.<region>.<domain>
Content-Length: <byte_size>
Content-Type: application/x-amz-json-1.1
Connection: Keep-Alive
X-Amz-Target: Kinesis_20131202.PutRecords
{
"Records": [
{
"Data": "<base64_encoded_event>",
"PartitionKey": "<partition_key>"
},
{
"Data": "<base64_encoded_event>",
"PartitionKey": "<partition_key>"
},
{
"Data": "<base64_encoded_event>",
"PartitionKey": "<partition_key>"
},
],
"StreamName": "<stream_name>"
}

How It Works

Authentication

Vector checks for AWS credentials in the following order:

  1. Environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY.

    ​2. The credential_process command in the AWS config file. (usually located at ~/.aws/config)

    ​3. The AWS credentials file. (usually located at ~/.aws/credentials)

  2. The ​IAM instance profile. (will only work if running on an EC2 instance with an instance profile/role)

If credentials are not found the healtcheck will fail and an error will be logged.

Obtaining an access key

In general, we recommend using instance profiles/roles whenever possible. In cases where this is not possible you can generate an AWS access key for any user within your AWS account. AWS provides a detailed guide on how to do this.

Buffers & Batches

The aws_kinesis_streams sink buffers & batches data as shown in the diagram above. You'll notice that Vector treats these concepts differently, instead of treating them as global concepts, Vector treats them as sink specific concepts. This isolates sinks, ensuring services disruptions are contained and delivery guarantees are honored.

Buffers types

The buffer.type option allows you to control buffer resource usage:

Type

Description

memory

Pros: Fast. Cons: Not persisted across restarts. Possible data loss in the event of a crash. Uses more memory.

disk

Pros: Persisted across restarts, durable. Uses much less memory. Cons: Slower, see below.

Buffer overflow

The buffer.when_full option allows you to control the behavior when the buffer overflows:

Type

Description

block

Applies back pressure until the buffer makes room. This will help to prevent data loss but will cause data to pile up on the edge.

drop_newest

Drops new data as it's received. This data is lost. This should be used when performance is the highest priority.

Batch flushing

Batches are flushed when 1 of 2 conditions are met:

  1. The batch age meets or exceeds the configured batch_timeout (default: 1 seconds).

  2. The batch size meets or exceeds the configured batch_size (default: 1049000 bytes).

Delivery Guarantee

This component offers an at least once delivery guarantee if your pipeline is configured to achieve this.

Encodings

The aws_kinesis_streams sink encodes events before writing them downstream. This is controlled via the encoding option which accepts the following options:

Encoding

Description

json

The payload will be encoded as a single JSON payload.

text

The payload will be encoded as new line delimited text, each line representing the value of the "message" key.

Dynamic encoding

By default, the encoding chosen is dynamic based on the explicit/implcit nature of the event's structure. For example, if this event is parsed (explicit structuring), Vector will use json to encode the structured data. If the event was not explicitly structured, the text encoding will be used.

To further explain why Vector adopts this default, take the simple example of accepting data over the tcp source and then connecting it directly to the aws_kinesis_streams sink. It is less surprising that the outgoing data reflects the incoming data exactly since it was not explicitly structured.

Environment Variables

Environment variables are supported through all of Vector's configuration. Simply add ${MY_ENV_VAR} in your Vector configuration file and the variable will be replaced before being evaluated.

You can learn more in the Environment Variables section.

Health Checks

Health checks ensure that the downstream service is accessible and ready to accept data. This check is performed upon sink initialization.

If the health check fails an error will be logged and Vector will proceed to start. If you'd like to exit immediately upon health check failure, you can pass the --require-healthy flag:

vector --config /etc/vector/vector.toml --require-healthy

And finally, if you'd like to disable health checks entirely for this sink you can set the healthcheck option to false.

Partitioning

By default, Vector issues random 16 byte values for each Kinesis record's partition key, evenly distributing records across your Kinesis partitions. Depending on your use case this might not be sufficient since random distribution does not preserve order. To override this, you can supply the partition_key_field option. This option represents a field on your event to use for the partition key value instead. This is useful if you have a field already on your event, and it also pairs nicely with the add_fields transform.

Missing keys or blank values

Kenisis requires a value for the partition key and therefore if the key is missing or the value is blank the event will be dropped and a warning level log event will be logged. As such, the field specified in the partition_key_field option should always contain a value.

Values that exceed 256 characters

If the value provided exceeds the maximum allowed length of 256 characters Vector will slice the value and use the first 256 characters.

Non-string values

Vector will coerce the value into a string.

Provisioning & capacity planning

This is generally outside the scope of Vector but worth touching on. When you supply your own partition key it opens up the possibility for "hot spots", and you should be aware of your data distribution for the key you're providing. Kinesis provides the ability to manually split shards to accomodate this. If they key you're using is dynamic and unpredictable we highly recommend recondsidering your ordering policy to allow for even and random distribution.

Rate Limits

Vector offers a few levers to control the rate and volume of requests to the downstream service. Start with the rate_limit_duration and rate_limit_num options to ensure Vector does not exceed the specified number of requests in the specified window. You can further control the pace at which this window is saturated with the request_in_flight_limit option, which will guarantee no more than the specified number of requests are in-flight at any given time.

Please note, Vector's defaults are carefully chosen and it should be rare that you need to adjust these. If you found a good reason to do so please share it with the Vector team by opening an issie.

Retry Policy

Vector will retry failed requests (status == 429, >= 500, and != 501). Other responses will not be retried. You can control the number of retry attempts and backoff rate with the retry_attempts and retry_backoff_secs options.

Timeouts

To ensure the pipeline does not halt when a service fails to respond Vector will abort requests after 30 seconds. This can be adjsuted with the request_timeout_secs option.

It is highly recommended that you do not lower value below the service's internal timeout, as this could create orphaned requests, pile on retries, and result in deuplicate data downstream.

Troubleshooting

The best place to start with troubleshooting is to check the Vector logs. This is typically located at /var/log/vector.log, then proceed to follow the Troubleshooting Guide.

If the Troubleshooting Guide does not resolve your issue, please:

  1. If encountered a bug, please file a bug report.

  2. If encountered a missing feature, please file a feature request.

  3. If you need help, join our chat/forum community. You can post a question and search previous questions.

Resources