Stato dell'orchestrazione personalizzato in Funzioni permanenti (Funzioni di Azure)
Articolo
Lo stato dell'orchestrazione personalizzato consente di impostare un valore di stato per la funzione dell'agente di orchestrazione. Questo stato viene fornito tramite l'API HTTP GetStatus o l'API SDK equivalente nell'oggetto client di orchestrazione.
Caso d'uso di esempio
Visualizzazione dello stato
I client possono eseguire il polling dell'endpoint di stato e visualizzare un elemento dell'interfaccia utente con lo stato che mostra la fase di esecuzione corrente. L'esempio seguente mostra la condivisione dello stato:
Questi esempi C# vengono scritti per Durable Functions 2.x e non sono compatibili con Durable Functions 1.x. Per altre informazioni sulle differenze tra le versioni, vedere l'articolo Durable Functions versioni.
Un altro scenario interessante è la segmentazione degli utenti con la restituzione di un output personalizzato in base a caratteristiche o interazioni univoche. Con l'aiuto dello stato dell'orchestrazione personalizzato, il codice lato client viene mantenuto generico. Tutte le modifiche principali avverranno sul lato server come illustrato nell'esempio seguente:
[FunctionName("CityRecommender")]
public static void Run(
[OrchestrationTrigger] IDurableOrchestrationContext context)
{
int userChoice = context.GetInput<int>();
switch (userChoice)
{
case 1:
context.SetCustomStatus(new
{
recommendedCities = new[] {"Tokyo", "Seattle"},
recommendedSeasons = new[] {"Spring", "Summer"}
});
break;
case 2:
context.SetCustomStatus(new
{
recommendedCities = new[] {"Seattle, London"},
recommendedSeasons = new[] {"Summer"}
});
break;
case 3:
context.SetCustomStatus(new
{
recommendedCities = new[] {"Tokyo, London"},
recommendedSeasons = new[] {"Spring", "Summer"}
});
break;
}
// Wait for user selection and refine the recommendation
}
CityRecommender Orchestrator
const df = require("durable-functions");
module.exports = df.orchestrator(function*(context) {
const userChoice = context.df.getInput();
switch (userChoice) {
case 1:
context.df.setCustomStatus({
recommendedCities: [ "Tokyo", "Seattle" ],
recommendedSeasons: [ "Spring", "Summer" ],
});
break;
case 2:
context.df.setCustomStatus({
recommendedCities: [ "Seattle", "London" ],
recommendedSeasons: [ "Summer" ],
});
break;
case 3:
context.df.setCustomStatus({
recommendedCity: [ "Tokyo", "London" ],
recommendedSeasons: [ "Spring", "Summer" ],
});
break;
}
// Wait for user selection and refine the recommendation
});
CityRecommender Orchestrator
import azure.functions as func
import azure.durable_functions as df
def orchestrator_function(context: df.DurableOrchestrationContext):
userChoice = int(context.get_input())
if userChoice == 1:
context.set_custom_status({
'recommendedCities' : ['Tokyo', 'Seattle'],
'recommendedSeasons' : ['Spring', 'Summer']
}))
if userChoice == 2:
context.set_custom_status({
'recommendedCities' : ['Seattle', 'London']
'recommendedSeasons' : ['Summer']
}))
if userChoice == 3:
context.set_custom_status({
'recommendedCities' : ['Tokyo', 'London'],
'recommendedSeasons' : ['Spring', 'Summer']
}))
# Wait for user selection and refine the recommendation
main = df.Orchestrator.create(orchestrator_function)
CityRecommender Orchestrator
param($Context)
$userChoice = $Context.Input -as [int]
if ($userChoice -eq 1) {
Set-DurableCustomStatus -CustomStatus @{ recommendedCities = @('Tokyo', 'Seattle');
recommendedSeasons = @('Spring', 'Summer')
}
}
if ($userChoice -eq 2) {
Set-DurableCustomStatus -CustomStatus @{ recommendedCities = @('Seattle', 'London');
recommendedSeasons = @('Summer')
}
}
if ($userChoice -eq 3) {
Set-DurableCustomStatus -CustomStatus @{ recommendedCities = @('Tokyo', 'London');
recommendedSeasons = @('Spring', 'Summer')
}
}
# Wait for user selection and refine the recommendation
@FunctionName("CityRecommender")
public void cityRecommender(
@DurableOrchestrationTrigger(name = "ctx") TaskOrchestrationContext ctx) {
int userChoice = ctx.getInput(int.class);
switch (userChoice) {
case 1:
ctx.setCustomStatus(new Recommendation(
new String[]{ "Tokyo", "Seattle" },
new String[]{ "Spring", "Summer" }));
break;
case 2:
ctx.setCustomStatus(new Recommendation(
new String[]{ "Seattle", "London" },
new String[]{ "Summer" }));
break;
case 3:
ctx.setCustomStatus(new Recommendation(
new String[]{ "Tokyo", "London" },
new String[]{ "Spring", "Summer" }));
break;
}
// Wait for user selection with an external event
}
class Recommendation {
public Recommendation() { }
public Recommendation(String[] cities, String[] seasons) {
this.recommendedCities = cities;
this.recommendedSeasons = seasons;
}
public String[] recommendedCities;
public String[] recommendedSeasons;
}
Specifiche delle istruzioni
L'agente di orchestrazione può offrire istruzioni univoche ai client tramite lo stato personalizzato. Le istruzioni dello stato personalizzate verranno mappate ai passaggi nel codice di orchestrazione:
[FunctionName("ReserveTicket")]
public static async Task<bool> Run(
[OrchestrationTrigger] IDurableOrchestrationContext context)
{
string userId = context.GetInput<string>();
int discount = await context.CallActivityAsync<int>("CalculateDiscount", userId);
context.SetCustomStatus(new
{
discount = discount,
discountTimeout = 60,
bookingUrl = "https://www.myawesomebookingweb.com",
});
bool isBookingConfirmed = await context.WaitForExternalEvent<bool>("BookingConfirmed");
context.SetCustomStatus(isBookingConfirmed
? new {message = "Thank you for confirming your booking."}
: new {message = "The booking was not confirmed on time. Please try again."});
return isBookingConfirmed;
}
const df = require("durable-functions");
module.exports = df.orchestrator(function*(context) {
const userId = context.df.getInput();
const discount = yield context.df.callActivity("CalculateDiscount", userId);
context.df.setCustomStatus({
discount,
discountTimeout = 60,
bookingUrl = "https://www.myawesomebookingweb.com",
});
const isBookingConfirmed = yield context.df.waitForExternalEvent("bookingConfirmed");
context.df.setCustomStatus(isBookingConfirmed
? { message: "Thank you for confirming your booking." }
: { message: "The booking was not confirmed on time. Please try again." }
);
return isBookingConfirmed;
});
import azure.functions as func
import azure.durable_functions as df
def orchestrator_function(context: df.DurableOrchestrationContext):
userId = int(context.get_input())
discount = yield context.call_activity('CalculateDiscount', userId)
status = { 'discount' : discount,
'discountTimeout' : 60,
'bookingUrl' : "https://www.myawesomebookingweb.com",
}
context.set_custom_status(status)
is_booking_confirmed = yield context.wait_for_external_event('BookingConfirmed')
context.set_custom_status({'message': 'Thank you for confirming your booking.'} if is_booking_confirmed
else {'message': 'The booking was not confirmed on time. Please try again.'})
return is_booking_confirmed
main = df.Orchestrator.create(orchestrator_function)
param($Context)
$userId = $Context.Input -as [int]
$discount = Invoke-DurableActivity -FunctionName 'CalculateDiscount' -Input $userId
$status = @{
discount = $discount;
discountTimeout = 60;
bookingUrl = "https://www.myawesomebookingweb.com"
}
Set-DurableCustomStatus -CustomStatus $status
$isBookingConfirmed = Invoke-DurableActivity -FunctionName 'BookingConfirmed'
if ($isBookingConfirmed) {
Set-DurableCustomStatus -CustomStatus @{message = 'Thank you for confirming your booking.'}
} else {
Set-DurableCustomStatus -CustomStatus @{message = 'The booking was not confirmed on time. Please try again.'}
}
return $isBookingConfirmed
@FunctionName("ReserveTicket")
public boolean reserveTicket(
@DurableOrchestrationTrigger(name = "ctx") TaskOrchestrationContext ctx) {
String userID = ctx.getInput(String.class);
int discount = ctx.callActivity("CalculateDiscount", userID, int.class).await();
ctx.setCustomStatus(new DiscountInfo(discount, 60, "https://www.myawesomebookingweb.com"));
boolean isConfirmed = ctx.waitForExternalEvent("BookingConfirmed", boolean.class).await();
if (isConfirmed) {
ctx.setCustomStatus("Thank you for confirming your booking.");
} else {
ctx.setCustomStatus("There was a problem confirming your booking. Please try again.");
}
return isConfirmed;
}
class DiscountInfo {
public DiscountInfo() { }
public DiscountInfo(int discount, int discountTimeout, String bookingUrl) {
this.discount = discount;
this.discountTimeout = discountTimeout;
this.bookingUrl = bookingUrl;
}
public int discount;
public int discountTimeout;
public String bookingUrl;
}
Esecuzione di query sullo stato personalizzato con HTTP
Nell'esempio seguente viene illustrato come eseguire query sui valori di stato personalizzati usando le API HTTP predefinite.
public static async Task SetStatusTest([OrchestrationTrigger] IDurableOrchestrationContext context)
{
// ...do work...
// update the status of the orchestration with some arbitrary data
var customStatus = new { nextActions = new [] {"A", "B", "C"}, foo = 2, };
context.SetCustomStatus(customStatus);
// ...do more work...
}
const df = require("durable-functions");
module.exports = df.orchestrator(function*(context) {
// ...do work...
// update the status of the orchestration with some arbitrary data
const customStatus = { nextActions: [ "A", "B", "C" ], foo: 2, };
context.df.setCustomStatus(customStatus);
// ...do more work...
});
import azure.functions as func
import azure.durable_functions as df
def orchestrator_function(context: df.DurableOrchestrationContext):
# ...do work...
custom_status = {'nextActions': ['A','B','C'], 'foo':2}
context.set_custom_status(custom_status)
# ...do more work...
main = df.Orchestrator.create(orchestrator_function)
@FunctionName("MyCustomStatusOrchestrator")
public void myCustomStatusOrchestrator(
@DurableOrchestrationTrigger(name = "ctx") TaskOrchestrationContext ctx) {
// ... do work ...
// update the status of the orchestration with some arbitrary data
CustomStatusPayload payload = new CustomStatusPayload();
payload.nextActions = new String[] { "A", "B", "C" };
payload.foo = 2;
ctx.setCustomStatus(payload);
// ... do more work ...
}
class CustomStatusPayload {
public String[] nextActions;
public int foo;
}
Mentre l'orchestrazione è in esecuzione, i client esterni possono recuperare questo stato personalizzato:
GET /runtime/webhooks/durabletask/instances/instance123
Il payload dello stato personalizzato è limitato a 16 KB di testo JSON UTF-16. È consigliabile usare l'archiviazione esterna se è necessario un payload più grande.