

App Functions SDK (Golang) - Beta

Welcome the App Functions SDK for EdgeX. This sdk is meant to provide all the plumbing necessary for developers to get started in processing/transforming/exporting data out of EdgeX.

Table of contents

• Getting Started
• Triggers
• Context API
• Built-In Functions
  • Filtering
  • Encryption
  • Conversion
  • Compressions
  • Core Data
  • Export Functions
• Configuration
• Error Handling
• Advanced Topics
  • Target Type
  • Command Line Options
  • Environment Variable Overrides
  • Store and Forward

Getting Started

Build Prerequisites

Please see the edgex-go README.

The SDK

The SDK is built around the idea of a "Functions Pipeline". A functions pipeline is a collection of various functions that process the data in the order that you've specified. The functions pipeline is executed by the specified trigger in the configuration.toml . The first function in the pipeline is called with the event that triggered the pipeline (ex. events.Model). Each successive call in the pipeline is called with the return result of the previous function. Let's take a look at a simple example that creates a pipeline to filter particular device ids and subsequently transform the data to XML:

package main

import (
    "fmt"
    "github.com/edgexfoundry/app-functions-sdk-go/appsdk"
    "github.com/edgexfoundry/app-functions-sdk-go/pkg/transforms"
    "os"
)

func main() {

    // 1) First thing to do is to create an instance of the EdgeX SDK, giving it a service key
    edgexSdk := &appsdk.AppFunctionsSDK{
        ServiceKey: "SimpleFilterXMLApp", // Key used by Registry (Aka Consul)
    }

    // 2) Next, we need to initialize the SDK
    if err := edgexSdk.Initialize(); err != nil {
        edgexSdk.LoggingClient.Error(fmt.Sprintf("SDK initialization failed: %v\n", err))
        os.Exit(-1)
    }

    // 3) Since our FilterByDeviceName Function requires the list of Device Names we would
    // like to search for, we'll go ahead and define that now.
    deviceNames := []string{"Random-Float-Device"}

    // 4) This is our pipeline configuration, the collection of functions to
    // execute every time an event is triggered.
    if err := edgexSdk.SetFunctionsPipeline(
            transforms.NewFilter(deviceNames).FilterByDeviceName, 
            transforms.NewConversion().TransformToXML,
        ); err != nil {
            edgexSdk.LoggingClient.Error(fmt.Sprintf("SDK SetPipeline failed: %v\n", err))
            os.Exit(-1)
        }

    // 5) shows how to access the application's specific configuration settings.
    appSettings := edgexSdk.ApplicationSettings()
    if appSettings != nil {
        appName, ok := appSettings["ApplicationName"]
        if ok {
            edgexSdk.LoggingClient.Info(fmt.Sprintf("%s now running...", appName))
        } else {
            edgexSdk.LoggingClient.Error("ApplicationName application setting not found")
            os.Exit(-1)
        }
    } else {
        edgexSdk.LoggingClient.Error("No application settings found")
        os.Exit(-1)
    }

    // 6) Lastly, we'll go ahead and tell the SDK to "start" and begin listening for events to trigger the pipeline.
    edgexSdk.MakeItRun()
}

The above example is meant to merely demonstrate the structure of your application. Notice that the output of the last function is not available anywhere inside this application. You must provide a function in order to work with the data from the previous function. Let's go ahead and add the following function that prints the output to the console.

func printXMLToConsole(edgexcontext *appcontext.Context, params ...interface{}) (bool,interface{}) {
  if len(params) < 1 { 
    // We didn't receive a result
    return false, errors.New("No Data Received")
  }
  println(params[0].(string))
  return true, nil
}

After placing the above function in your code, the next step is to modify the pipeline to call this function:

edgexSdk.SetFunctionsPipeline(
  transforms.NewFilter(deviceNames).FilterByDeviceName, 
  transforms.NewConversion().TransformToXML,
  printXMLToConsole //notice this is not a function call, but simply a function pointer. 
)

After making the above modifications, you should now see data printing out to the console in XML when an event is triggered. > You can find this example in the /examples directory located in this repository. You can also use the provided EdgeX Applications Function SDK.postman\_collection.json file to load into postman to trigger the sample pipeline.

Up until this point, the pipeline has been triggered by an event over HTTP and the data at the end of that pipeline lands in the last function specified. In the example, data ends up printed to the console. Perhaps we'd like to send the data back to where it came from. In the case of an HTTP trigger, this would be the HTTP response. In the case of a message bus, this could be a new topic to send the data back to for other applications that wish to receive it. To do this, simply call edgexcontext.Complete([]byte outputData) passing in the data you wish to "respond" with. In the above printXMLToConsole(...) function, replace println(params[0].(string)) with edgexcontext.Complete([]byte(params[0].(string))). You should now see the response in your postman window when testing the pipeline.

Examples

The App Service Examples repo contains a variety of simple to advanced example Application Services built upon the App Functions SDK. Examples that once were in the examples folder of the SDK have been moved to this Examples repo.

Triggers

Triggers determine how the app functions pipeline begins execution. In the simple example provided above, an HTTP trigger is used. The trigger is determine by the configuration.toml file located in the /res directory under a section called [Binding]. Check out the Configuration Section for more information about the toml file.

Message Bus Trigger

A message bus trigger will execute the pipeline every time data is received off of the configured topic.

Type and Topic configuration

Here's an example:

Type="messagebus" 
SubscribeTopic="events"
PublishTopic=""

The Type= is set to "messagebus". EdgeX Core Data is publishing data to the events topic. So to receive data from core data, you can set your SubscribeTopic= either to "" or "events". You may also designate a PublishTopic= if you wish to publish data back to the message bus. edgexcontext.Complete([]byte outputData) - Will send data back to back to the message bus with the topic specified in the PublishTopic= property #### Message bus connection configuration The other piece of configuration required are the connection settings:

[MessageBus]
Type = 'zero' #specifies of message bus (i.e zero for ZMQ)
    [MessageBus.PublishHost]
        Host = '*'
        Port = 5564
        Protocol = 'tcp'
    [MessageBus.SubscribeHost]
        Host = 'localhost'
        Port = 5563
        Protocol = 'tcp'

By default, EdgeX Core Data publishes data to the events topic on port 5563. The publish host is used if publishing data back to the message bus. >Important Note: Publish Host MUST be different for every topic you wish to publish to since the SDK will bind to the specific port. 5563 for example cannot be used to publish since EdgeX Core Data has bound to that port. Similarly, you cannot have two separate instances of the app functions SDK running publishing to the same port.

HTTP Trigger

Designating an HTTP trigger will allow the pipeline to be triggered by a RESTful POST call to http://[host]:[port]/trigger/. The body of the POST must be an EdgeX event.

edgexcontext.Complete([]byte outputData) - Will send the specified data as the response to the request that originally triggered the HTTP Request.

Context API

The context parameter passed to each function/transform provides operations and data associated with each execution of the pipeline. Let's take a look at a few of the properties that are available:

type Context struct {
    // ID of the EdgeX Event -- will be filled for a received JSON Event
    EventID string

    // Checksum of the EdgeX Event -- will be filled for a received CBOR Event
    EventChecksum string

    // This is the ID used to track the EdgeX event through entire EdgeX framework.
    CorrelationID string

    // OutputData is used for specifying the data that is to be outputted. Leverage the .Complete() function to set.
    OutputData []byte

    // This holds the configuration for your service. This is the preferred way to access your custom application settings that have been set in the configuration. 
    Configuration common.ConfigurationStruct

    // LoggingClient is exposed to allow logging following the preferred logging strategy within EdgeX.
    LoggingClient logger.LoggingClient

    // EventClient exposes Core Data's EventClient API
    EventClient coredata.EventClient

    // ValueDescriptorClient exposes Core Data's ValueDescriptor API
    ValueDescriptorClient coredata.ValueDescriptorClient

    // CommandClient exposes Core Commands's Command API
    CommandClient command.CommandClient

    // NotificationsClient exposes Support Notification's Notifications API
    NotificationsClient notifications.NotificationsClient

    // RetryData holds the data to be stored for later retry when the pipeline function returns an error
    RetryData []byte
}

LoggingClient

The LoggingClient exposed on the context is available to leverage logging libraries/service utilized throughout the EdgeX framework. The SDK has initialized everything so it can be used to log Trace, Debug, Warn, Info, and Error messages as appropriate. See examples/simple-filter-xml/main.go for an example of how to use the LoggingClient.

EventClient

The EventClient exposed on the context is available to leverage Core Data's Event API. See interface definition for more details. This client is useful for querying events and is used by the MarkAsPushed convenience API described below.

ValueDescriptorClient

The ValueDescriptorClient exposed on the context is available to leverage Core Data's ValueDescriptor API. See interface definition for more details. Useful for looking up the value descriptor for a reading received.

CommandClient

The CommandClient exposed on the context is available to leverage Core Command's Command API. See interface definition for more details. Useful for sending commands to devices.

NotificationsClient

The CommandClient exposed on the context is available to leverage Support Notifications' Notifications API. See README for more details. Useful for sending notifications.

Note about Clients

Each of the clients above is only initialized if the Clients section of the configuration contains an entry for the service associated with the Client API. If it isn't in the configuration the client will be nil. Your code must check for nil to avoid panic in case it is missing from the configuration. Only add the clients to your configuration that your Application Service will actually be using. All application services need the Logging and many will need Core-Data. The following is an example Clients section of a configuration.toml with all supported clients specified:

[Clients]
  [Clients.Logging]
  Protocol = "http"
  Host = "localhost"
  Port = 48061

  [Clients.CoreData]
  Protocol = 'http'
  Host = 'localhost'
  Port = 48080

  [Clients.Command]
  Protocol = 'http'
  Host = 'localhost'
  Port = 48082

  [Clients.Notifications]
  Protocol = 'http'
  Host = 'localhost'
  Port = 48060

.MarkAsPushed()

.MarkAsPushed() is used to indicate to EdgeX Core Data that an event has been "pushed" and is no longer required to be stored. The scheduler service will purge all events that have been marked as pushed based on the configured schedule. By default, it is once daily at midnight. If you leverage the built in export functions (i.e. HTTP Export, or MQTT Export), then the event will automatically be marked as pushed upon a successful export. ### .PushToCore() .PushToCore(string deviceName, string readingName, byte[] value) is used to push data to EdgeX Core Data so that it can be shared with other applications that are subscribed to the message bus that core-data publishes to. deviceName can be set as you like along with the readingName which will be set on the EdgeX event sent to CoreData. This function will return the new EdgeX Event with the ID populated, however the CorrelationId will not be available.

NOTE: If validation is turned on in CoreServices then your deviceName and readingName must exist in the CoreMetadata and be properly registered in EdgeX.

.SetRetryData()

.SetRetryData(payload []byte) can be used to store data for later retry. This is useful when creating a custom export function that needs to retry on failure sending the data. The payload data will be stored for later retry based on Store and Forward configuration. When the retry is triggered, the function pipeline will be re-executed starting with the function that called this API. That function will be passed the stored data, so it is important that all transformations occur in functions prior to the export function. The Context will also be restored to the state when the function called this API. See Store and Forward for more details.

NOTE: Store and Forward be must enabled when calling this API.

Built-In Transforms/Functions

All transforms define a type and a New function which is used to initialize an instance of the type with the required parameters. These instances returned by these New functions give access to their appropriate pipeline function pointers when build the function pipeline.

E.G. NewFilter([] {"Device1", "Device2"}).FilterByDeviceName

Filtering

There are two basic types of filtering included in the SDK to add to your pipeline. Theses provided Filter functions return a type of events.Model. If filtering results in no remaining data, the pipeline execution for that pass is terminated. If no values are provided for filtering, then data flows through unfiltered. -NewFilter([]string filterValues) - This function returns a Filter instance initialized with the passed in filter values. This Filter instance is used to access the following filter functions that will operate using the specified filter values. -FilterByDeviceName - This function will filter the event data down to the specified device names and return the filtered data to the pipeline. - FilterByValueDescriptor - This function will filter the event data down to the specified device value descriptor and return the filtered data to the pipeline.

Encryption

There is one encryption transform included in the SDK that can be added to your pipeline.

• NewEncryption(key string, initializationVector string) - This function returns a Encryption instance initialized with the passed in key and initialization vector. This Encryption instance is used to access the following encryption function that will use the specified key and initialization vector.
  • EncryptWithAES - This function receives a either a string, []byte, or json.Marshaller type and encrypts it using AES encryption and returns a []byte to the pipeline.

Conversion

There are two conversions included in the SDK that can be added to your pipeline. These transforms return a string.

• NewConversion() - This function returns a Conversion instance that is used to access the following conversion functions:
  • TransformToXML - This function receives an events.Model type, converts it to XML format and returns the XML string to the pipeline.
  • TransformToJSON - This function receives an events.Model type and converts it to JSON format and returns the JSON string to the pipeline.

### Compressions There are two compression types included in the SDK that can be added to your pipeline. These transforms return a []byte.

• NewCompression() - This function returns a Compression instance that is used to access the following compression functions:
  • CompressWithGZIP - This function receives either a string,[]byte, or json.Marshaler type, GZIP compresses the data, converts result to base64 encoded string, which is returned as a []byte to the pipeline.
  • CompressWithZLIB - This function receives either a string,[]byte, or json.Marshaler type, ZLIB compresses the data, converts result to base64 encoded string, which is returned as a []byte to the pipeline.

CoreData Functions

These are functions that enable interactions with the CoreData REST API. - NewCoreData() - This function returns a CoreData instance. This CoreData instance is used to access the following function(s). - MarkAsPushed - This function provides the MarkAsPushed function from the context as a First-Class Transform that can be called in your pipeline. See Definition Above. The data passed into this function from the pipeline is passed along unmodifed since all required information is provided on the context (EventId, CorrelationId,etc.. ) - PushToCore - This function provides the PushToCore function from the context as a First-Class Transform that can be called in your pipeline. See Definition Above. The data passed into this function from the pipeline is wrapped in an EdgeX event with the deviceName and readingName that were set upon instantiation and then sent to CoreData to be added as an event. Returns the new EdgeX event with ID populated.

> NOTE: If validation is turned on in CoreServices then your `deviceName` and `readingName` must exist in the CoreMetadata and be properly registered in EdgeX.

Export Functions

There are two export functions included in the SDK that can be added to your pipeline. -NewHTTPSender(url string, mimeType string, persistOnError bool) -This function returns a HTTPSender instance initialized with the passed in url, mime type and persistOnError values. This HTTPSender instance is used to access the following functions that will use the required url and optional mime type and persistOnError:

• HTTPPost - This function receives either a string,[]byte, or json.Marshaler type from the previous function in the pipeline and posts it to the configured endpoint. If no previous function exists, then the event that triggered the pipeline, marshaled to json, will be used. Currently, only unauthenticated endpoints are supported. Authenticated endpoints will be supported in the future. If the post fails and persistOnErroris true and Store and Forward is enabled, the data will be stored for later retry. See Store and Forward for more details

• NewMQTTSender(logging logger.LoggingClient, addr models.Addressable, keyCertPair *KeyCertPair, mqttConfig MqttConfig, persistOnError bool)

  - This function returns a MQTTSender instance initialized with the passed in MQTT configuration . This MQTTSender instance is used to access the following function that will use the specified MQTT configuration

  • KeyCertPair - This structure holds the Key and Certificate information for when using secure TLS connection to the broker. Can be nil if not using secure TLS connection.

  • MqttConfig - This structure holds addition MQTT configuration settings.

    Qos            byte
    Retain         bool
    AutoReconnect  bool
    SkipCertVerify bool
    User           string
    Password       string
    

    The GO complier will default these to 0, false and "", so you only need to set the fields that your usage requires that differ from the default.

  • MQTTSend - This function receives either a string,[]byte, or json.Marshaler type from the previous function in the pipeline and sends it to the specified MQTT broker. If no previous function exists, then the event that triggered the pipeline, marshaled to json, will be used. If the send fails and persistOnErroris true and Store and Forward is enabled, the data will be stored for later retry. See Store and Forward for more details

Output Functions

There is one output function included in the SDK that can be added to your pipeline.

• NewOutput() - This function returns a Output instance that is used to access the following output function:
  • SetOutput - This function receives either a string,[]byte, or json.Marshaler type from the previous function in the pipeline and sets it as the output data for the pipeline to return to the configured trigger. If configured to use message bus, the data will be published to the message bus as determined by the MessageBus and Binding configuration. If configured to use HTTP trigger the data is returned as the HTTP response. Note that calling Complete() from the Context API in a custom function can be used in place of adding this function to your pipeline

Configuration

Similar to other EdgeX services, configuration is first determined by the configuration.toml file in the /res folder. If -r is passed to the application on startup, the SDK will leverage the provided registry (i.e Consul) to push configuration from the file into the registry and monitor configuration from there. You will find the configuration under the edgex/appservices/1.0/ key. There are two primary sections in the configuration.toml file that will need to be set that are specific to the AppFunctionsSDK. 1) [Binding] - This specifies the trigger type and associated data required to configure a trigger.

[Binding]
Type=""
SubscribeTopic=""
PublishTopic=""

2) [ApplicationSettings] - Is used for custom application settings and is accessed via the ApplicationSettings() API. The ApplicationSettings API returns a map[string] string containing the contents on the ApplicationSetting section of the configuration.toml file.

[ApplicationSettings]
ApplicationName = "My Application Service"

Error Handling

• Each transform returns a true or false as part of the return signature. This is called the continuePipeline flag and indicates whether the SDK should continue calling successive transforms in the pipeline.
• return false, nil will stop the pipeline and stop processing the event. This is useful for example when filtering on values and nothing matches the criteria you've filtered on.
• return false, error, will stop the pipeline as well and the SDK will log the errorString you have returned.
• Returning true tells the SDK to continue, and will call the next function in the pipeline with your result.
• The SDK will return control back to main when receiving a SIGTERM/SIGINT event to allow for custom clean up.

Advanced Topics

The following items discuss topics that are a bit beyond the basic use cases of the Application Functions SDK when interacting with EdgeX.

Configurable Functions Pipeline

This SDK provides the capability to define the functions pipeline via configuration rather than code using the app-service-configurable application service. See app-service-configurable README for more details.

Using The Webserver

It is not uncommon to require your own API endpoints when building an app service. Rather than spin up your own webserver inside of your app (alongside the already existing running webserver), we've exposed a method that allows you add your own routes to the existing webserver. A few routes are reserved and cannot be used: - /api/version -/api/v1/ping - /api/v1/metrics - /api/v1/config - /api/v1/trigger To add your own route, use the AddRoute(route string, handler func(nethttp.ResponseWriter, *nethttp.Request), methods ...string) function provided on the sdk. Here's an example:

edgexSdk.AddRoute("/myroute", func(writer http.ResponseWriter, req *http.Request) {
    context := req.Context().Value(appsdk.SDKKey).(*appsdk.AppFunctionsSDK) 
        context.LoggingClient.Info("TEST") // alternative to edgexSdk.LoggingClient.Info("TEST")
        writer.Header().Set("Content-Type", "text/plain")
        writer.Write([]byte("hello"))
        writer.WriteHeader(200)
}, "GET")

Under the hood, this simply adds the provided route, handler, and method to the gorilla mux.Router we use in the SDK. For more information you can check out the github repo here. You can access the resources such as the logging client by accessing the context as shown above -- this is useful for when your routes might not be defined in your main.go where you have access to the edgexSdk instance.

Target Type

The target type is the object type of the incoming data that is sent to the first function in the function pipeline. By default this is an EdgeX Event since typical usage is receiving events from Core Data via Message Bus.

For other usages where the data is not events coming from Core Data, the TargetType of the accepted incoming data can be set when the SDK instance is created. There are scenarios where the incoming data is not an EdgeX Event. One example scenario is 2 application services are chained via the Message Bus. The output of the first service back to the Messages Bus is inference data from analyzing the original input Eventdata. The second service needs to be able to let the SDK know the target type of the input data it is expecting.

For usages where the incoming data is not events, the TargetType of the excepted incoming data can the set when the SDK instance is created.

Example:

type Person struct {
    FirstName string `json:"first_name"`
    LastName  string `json:"last_name"`
}

edgexSdk := &appsdk.AppFunctionsSDK {
    ServiceKey: serviceKey, 
    TargetType: &Person{},
}

Note that TargetType must be set to a pointer to an instance of your target type such as &Person{} . The first function in your function pipeline will be passed an instance of your target type, not a pointer to it. In the example above the first function in the pipeline would start something like:

func MyPersonFunction(edgexcontext *appcontext.Context, params ...interface{}) (bool, interface{}) {

    edgexcontext.LoggingClient.Debug("MyPersonFunction")

    if len(params) < 1 {
        // We didn't receive a result
        return false, nil
    }

    person, ok := params[0].(Person)
    if !ok {
        return false, errors.New("type received is not a Person")
    }

    ....

The SDK supports unmarshaling JSON or CBOR encoded data into an instance of the target type. If your incoming data is not JSON or CBOR encoded, you then need to set the TargetType to &[]byte.

If the target type is set to &[]byte the incoming data will not be unmarshaled. The content type, if set, will be passed as the second parameter to the first function in your pipeline. Your first function will be responsible for decoding the data or not.

Command Line Options

The following command line options are available

-c=<path>
--confdir=<path>
      Specify an alternate configuration directory.
-p=<profile>
--profile=<profile>
      Specify a profile other than default.
-r    
--registry
      Indicates the service should use the registry.
-o    
-overwrite
      Overwrite configuration in the Registry with local values.
-s    
-skipVersionCheck
      Indicates the service should skip the Core Service's version compatibility check.

Examples:

simple-filter-xml -r -c=./res -p=docker

or

simple-filter-xml --registry --confdir=./res --profile=docker

Environment Variable Overrides

All the configuration settings from the configuration.toml file can be overridden by environment variables. Except for two special cases listed below, the overrides only occur when the configuration values are first pushed into the Registry. Once the values are in the Registry, the Registry values are always used.

The environment variable names have the following format:

<TOML Key>
<TOML Section>_<TOML Key>
<TOML Section>_<TOML Sub-Section>_<TOML Key>

Examples:

TOML   : FailLimit = 30
ENVVAR : FailLimit=100

TOML   : [Logging]
         EnableRemote = false
ENVVAR : Logging.EnableRemote=true

TOML   : [Clients]
            [Clients.CoreData]
            Host = 'localhost'
ENVVAR : Clients_CoreData_Host=edgex-core-data

edgex_registry

This environment variable overrides the Registry connection information and occurs every time the application service starts. The value is in the format of a URL.

edgex_registry=consul://edgex-core-consul:8500

This sets the Registry information fields as follows:
    Type: consul
    Host: edgex-core-consul
    Port: 8500

edgex_service

This environment variable overrides the Service connection information and occurs every time the application service starts. The value is in the format of a URL.

edgex_service=http://192.168.1.2:4903

This sets the Service information fields as follows:
    Protocol: http
    Host: 192.168.1.2
    Port: 4903

edgex_profile

This environment variable overrides the command line profile argument. It will replace the current value passed via the -p or --profile, if one exists. If not specified it will add the --profile argument. This is useful when running the service via docker-compose.

Using docker-compose:

app-service-configurable-rules:
  image: edgexfoundry/docker-app-service-configurable:1.1.0
  environment: 
    - edgex_profile : "rules-engine"
  ports:
    - "48095:48095"
  container_name: edgex-app-service-configurable
  hostname: edgex-app-service-configurable
  networks:
    edgex-network:
      aliases:
        - edgex-app-service-configurable
  depends_on:
    - data
    - command

This sets the --profile=docker-rules-engine command line argument so that the application service uses the docker-rules-engine configuration profile which resides at /res/docker-rules-engine/configuration.toml

Note that Application Services no longer use docker profiles. They use Environment Overrides in the docker compose file to make the necessary changes to the configuration for running in Docker. See theEnvironment Variable Overrides For Dockersection inApp Service Configurable's README for more details and an example.

Store and Forward

The Store and Forward capability allows for export functions to persist data on failure and for the export of the data to be retried at a later time.

Note: The order the data exported via this retry mechanism is not guaranteed to be the same order in which the data was initial received from Core Data

Configuration

Two sections of configuration have been added for Store and Forward.

Writable.StoreAndForward allows enabling, setting the interval between retries and the max number of retries. If running with Registry, these setting can be changed on the fly without having to restart the service.

[Writable.StoreAndForward]
  Enabled = false
  RetryInterval = '5m'
  MaxRetryCount = 10

Database describes which database type to use, mongodb or redisdb, and the information required to connect to the database. This section is required if Store and Forward is enabled, otherwise it is currently optional.

[Database]
Type = "mongodb"
Host = "localhost"
Port = 27017
Timeout = '5s'
Username = ""
Password = ""

How it works

When an export function encounters an error sending data it can call SetRetryData(payload []byte) on the Context. This will store the data for later retry. If the application service is stop and then restarted while stored data hasn't been successfully exported, the export retry will resume once the service is up and running again.

Note: It is important that export functions return an error and stop pipeline execution after the call to SetRetryData. See HTTPPost function in SDK as an example

When the RetryInterval expires, the function pipeline will be re-executed starting with the export function that saved the data. The saved data will be passed to the export function which can then attempt to resend the data.

NOTE: The export function will receive the data as it was stored, so it is important that any transformation of the data occur in functions prior to the export function. The export function should only export the data that it receives.

One of three out comes can occur after the export retried has completed.

1. Export retry was successful

   In this case the stored data is removed from the database and the execution of the pipeline functions after the export function, if any, continues.

2. Export retry fails and retry count has not been exceeded

   In this case the store data is updated in the database with the incremented retry count

3. Export retry fails and retry count has been exceeded

   In this case the store data is removed from the database and never retried again.