Reference

• 1: Horreum Clients
• 2: Webhook Reference
• 3: API Filter Queries
• 4: Troubleshooting Functions

1 - Horreum Clients

Starting from version 0.13, Horreum comes with some autogenerated clients out of the box that can be used to interact with the Horreum server programmatically.

Right now there are two clients, one for Python and the other one for Go.

Both clients are autogenerated using Kiota openapi generator, and at the moment of writing the autogenerated client is directly exposed through the HorreumClient instance. Therefore, the interaction with the raw client follows the Kiota experience, more details can be found in the Kiota documentation.

Prerequisites

In the below sections you can find some usage examples for both clients, the only precondition is that you must have Horreum server up and running (and accessible).

You can find more details on how to setup Horreum server in the Get Started section.

Python Client

Install the latest Horreum client

pip install horreum

Import required Python libraries

import asyncio

asyncio is required because the client leverages it to perform async requests to the server.

from horreum import new_horreum_client, HorreumCredentials

new_horreum_client utility function to setup the HorreumClient instance

Initialize the Horreum client

client = await new_horreum_client(base_url="http://localhost:8080", credentials=HorreumCredentials(username=username, password=password))

Now let’s start playing with it

# sync call
print("Client version:", client.version())

# async call
print("Server version:", (await client.raw_client.api.config.version.get()).version)

As you might have noticed, the Kiota api follows the REST url pattern. The previous async call was the equivalent of doing curl -X 'GET' 'http://localhost:8080/api/config/version'. The URL pattern api/config/version is replaced by dot notation api.config.version.<HTTP_METHOD>.

Golang Client

Install the latest Horreum client

go get github.com/hyperfoil/horreum-client-golang

Import required Go libraries

import (
    "github.com/hyperfoil/horreum-client-golang/pkg/horreum"
)

Define username and password

var (
	username = "user"
	password = "secret"
)

Initialize the Horreum client

client, err := horreum.NewHorreumClient("http://localhost:8080",
		&horreum.HorreumCredentials{
			Username: &username,
			Password: &password,
		},
		nil,
	)
if err != nil {
    log.Fatalf("error creating Horreum client: %s", err.Error())
}

Now let’s start playing with it

// sync call
fmt.Printf("Server version: %s\n", *v.GetVersion())

// async call
v, err := client.RawClient.Api().Config().Version().Get(context.Background(), nil)
if err != nil {
    log.Fatalf("error getting server version: %s", err.Error())
    os.Exit(1)
}

Similarly to the Python example, the previous async call was the equivalent of doing curl -X 'GET' 'http://localhost:8080/api/config/version'. The URL pattern api/config/version is replaced by method chain invocation api().config().version().<HTTP_METHOD>. Here you need to provide your own context to the actual HTTP method call.

2 - Webhook Reference

3 - API Filter Queries

Some APIs provide the ability to filter results when HTTP API response content type is application-json. The API Filter Query operates on the server side to reduce payload size.

API operations that support Filter Query provide an HTTP request parameter named query. The value of the query parameter is a JSONPath expression. Any syntax errors in the JSONPath will cause the operation to fail with details of the syntax issue.

To use the Filter Query with an operation that returns a Run data can be done as follows.

curl -s  'http://localhost:8080/api/sql/1/queryrun?query=$.*&array=true'  -H 'content-type: application/json' -H 'Authorization: Bearer '$TOKEN
{
  "valid": true,
  "jsonpath": "$.*",
  "errorCode": 0,
  "sqlState": null,
  "reason": null,
  "sql": null,
  "value": "[\"urn:acme:benchmark:0.1\", [{\"test\": \"Foo\", \"duration\": 10, \"requests\": 123}, {\"test\": \"Bar\", \"duration\": 20, \"requests\": 456}], \"defec8eddeadbeafcafebabeb16b00b5\", \"This gets lost by the transformer\"]}"
}

The response contains an enclosing JSON object that is common for Filter Query operations. The value property contains an array. The elements of the array contain the values of properties in the uploaded Run JSON. Note the additional HTTP request query parameter named array. Setting array is necessary to return dynamically sized results.

To filter the above API operation to retrieve only the results property of the uploaded Run object the query parameter is defined as $.results

curl -s  'http://localhost:8080/api/sql/1/queryrun?query=$.results&array=true'  -H 'content-type: application/json' -H 'Authorization: Bearer '$TOKEN
{
  "valid": true,
  "jsonpath": "$.results",
  "errorCode": 0,
  "sqlState": null,
  "reason": null,
  "sql": null,
  "value": "[[{\"test\": \"Foo\", \"duration\": 10, \"requests\": 123}, {\"test\": \"Bar\", \"duration\": 20, \"requests\": 456}]]"
}

The response JSON object this time contains the results property value JSON.

4 - Troubleshooting Functions

Troubleshooting Functions and Combination Function

Creating Scheam Labels is one of the key capabilities that enables working with loosely defined data. Using a JSON Validation Schema provides useful guarentees data in an uploaded Run JSON conforms to expected data types. A Validation Schema is optional but when not used makes the handling of metrics unreliable as data types may change.

To troubleshoot use of JavaScript Function the following techniques can be used

• The typeof function to discover the data-type of the Extractor variable, Object property or Array element in the Label Combination Function, using the Test label calculation button to reveal the JavaScript data-type

• View the Labels calculation Log window and display INFO or DEBUG log level, this reveals syntax errors when the function is executed

• Use the Show label values button, provides a quick visual check of resulting values for the Run Dataset

Effective label values window reveals the anticipated value or error text which can be due to the following:

• Nan - Numerical expression involves a data-type that is incompatible
• (undefined) - Incorrect JSONPath field value used in the Label Extractor
• empty - Absence of any value in the uploaded Run