RAP

The random avro producer

RAP is a fast Avro generator + Kafka producer that allows to fully control the data generation via a simple yaml configuration.

Getting started

The quickest way to try RAP is to run it against a local single node Kafka cluster.

Download the latest release of RAP and the example config file

RAP_VERSION=0.1.1 bash -c 'curl -Lo rap_${RAP_VERSION}.tar.gz \
https://github.com/andrewinci/rap/releases/download/v${RAP_VERSION}/rap_${RAP_VERSION}_$(uname)_$(uname -m).tar.gz && \
tar xvf rap_${RAP_VERSION}.tar.gz'
curl -o config.yaml https://raw.githubusercontent.com/andrewinci/rap/main/example/local_cluster.yaml

Start a local cluster with the RedPanda image

docker run -d --pull=always --name=redpanda-1 --rm \
  -p 9092:9092 \
  -p 9644:9644 \
  docker.vectorized.io/vectorized/redpanda:latest \
  redpanda start \
  --overprovisioned \
  --smp 1  \
  --memory 1G \
  --reserve-memory 0M \
  --node-id 0 \
  --check=false

Generate 2M records with RAP
```
./rap config.yaml
```
(optional) Verify the content of the topic with Insulator

Configuration

Use a .yaml file to configure the avro generation. Here an example config file with all the options:

kafka:
  clusterEndpoint: exampleEndpoint # the kafka endpoint (can also be passed with an env variable like $KAFKA_ENDPOINT)
  security: sasl #one of: none, sasl, mtls is expected here
  # the schema registry is optional and only required if schemaName is used in a producer
  schemaRegistry:
    # the schema registry endpoint (can also be passed with an env variable like $SR_ENDPOINT)
    endpoint: schemaRegistryEndpoint 
    username: $SR_USER
    password: $SR_PASSWORD
    
  sasl:
    # retrieve username and password from an env variable
    username: $KAFKA_USERNAME 
    password: $KAFKA_PASSWORD
  # mtls: todo

producers:
  - name: producer1 # an identifier for this producer
    numberOfMessages: 2000  # number of messages to generate from this producer
    topic: mytest-topic
    avro:
      schema: 
        id:     # the id registered in the schema registry
        raw: {} # the avro schema in json format
      generationRules: # set of rules to configure the generation of specific fields
        key: keyGen # special generation rule used to generate the record key
        .Name: nameGen 
        .SubRecord.Email: emailGen
        .SubRecord.Age: ageGen
      generators: # map of generators that can be used in the rules 
        nameGen: "{string}[Mary|James|Patricia|Robert]{1}"
        emailGen: "{string}[a-z]{10}[@]{1}[a-z]{10}[.org|.com]{1}"
        ageGen: "{int}[2|3|4|5|6]{1}[0-9]{1}"
        keyGen: "{string}[a-Z | 0-9]{10}"
  - name: producer2
    numberOfMessages: 20000
    topic: mytest-topic
    avro:
        schemaName: test-value # retrieve the schema from schema registry
        generationRules: # set of rules to configure the generation of specific fields
            key: keyGen # special generation rule used to generate the record key
        generators: # map of generators that can be used in the rules 
            keyGen: "{string}[0-9]{10}"
...

NOTE: Only the Kafka configurations can be passed via env variables

Generation rules

The generation rules describe how a specific field needs to be generated. The key of the yaml map identify what schema entity we need to target for the generation while the value is the generator to use.

The key can be:

a path to a field of the schema
an avro type: boolean int long float double bytes string
the value key to specify how to generate the key of the Kafka record

The priority of the generators is:

field path generator from config
avro type generator from config
default type generator

Schema field path

For example, .f1.f2 identify the field f2 nested in the record at field f1 which is part of the root record.

To describe a path through an avro union type, it is necessary to specify which type of the Union we want to follow.

For example, in the following schema fragment the path to f2 is .f1.Nested.f2. This path also tell to the avroGen to always pick the Nested side of the Union, therefore f1 will never be set as a string value.

{
  "type": "record",
  "fields": [
    {
      "name": "f1",
      "type": [
        "string",
        {
          "type": "record",
          "name": "Nested",
          "fields": [
            { "name": "f2", "type": "int"}

To describe a path to an element of an array follow the same rules as nested object.

For example, in the schema fragment below the path to stringField is simply .testField.stringField.

{
  "type": "record",
  "fields": [
    {
      "name": "testField",
      "type": {
        "type": "array",
        "items" : {
          "type" : "record",
          "name" : "ArrayObj",
          "fields" : [
            { "name": "stringField", "type": "string" }

For array, it is possible to specify the length of the generated arrays postponing .len() to the array path.

For example, in the schema above, we could use the following generator rule to restrict the len of the testField to one of 0,1,2.

generationRules:
  .testField.len(): lenGen
generators:
  lenGen: "{int}[0|1|2]{1}"

Generators syntax

To customize the generation of the fields it is possible to provide a pattern. The generic structure of a data gen pattern is:

{type}[content-restrictions]{count}[content-restrictions2]{count2}....

Where type can be one of the following avro type:
boolean int long float double bytes string

content-restriction can be:

an interval: a-z A-Z a-Z 0-9
a constant value: testvalue
a function: uuid() timestamp_ms()
a combination of intervals and constants: a-z | 0-9 | test

The field count tells the generator how many times the generation should be performed accordingly to the content-restriction. The result of each generation is concatenated.

Note for Avro enums use the {string} type generator making sure that the output matches one of the symbols.

Examples

Generate a constant value
{string}[test]{1} will always generate the constant value test

Generate alphanumeric string
{string}[a-Z|0-9]{10} will generate a random alphanumeric string of length 10

Generate a random email
{string}[a-z]{10}[@]{1}[a-z]{10}[.org|.com]{1}

Generate a random number
{int}[0-9]{5} will generate a random number of 5 digits.

Generate a random v4 uuid
{string}[uuid()]{1}

Development

Run tests with go test ./...

Credits

https://docs.redpanda.com/docs/quickstart/quick-start-docker/

TODO:

support logical types in field gen
support mtls authentication
support split yaml file
docker image and helm chart

andrewinci / rap

RAP