ReTry & ReTryWhen Operators help us to retry a failed observable in Angular. These operators are useful in Error handling. They both resubscribe to the source observable when they receive onError() notification.
WThe ReTry Angule RxJs operator retries a failed source observable count number of times. If the count is not provided then it tries indefinitely
Syntax
retry<T>(count: number = -1): MonoTypeOperatorFunction<T>
Mirrors the source observable, if there are no errors.
If the source observable calls error notification, it does not propagate it. But re subscribes to the source again for a maximum of count times.
The Retry operator retries the operation immediately
In the example below, map operator throws an error if the source emits a value greater than 3.
The ReTry(2) retries the operation twice before erroring out.
import { Component } from "@angular/core";
import { map, retry, tap } from "rxjs/operators";
import { interval } from "rxjs";
@Component({
selector: "my-app",
template: `
<h1>Retry And ReTryWhen Example</h1>
`,
styleUrls: ["./app.component.css"]
})
export class AppComponent {
ngOnInit() {
interval(1000)
.pipe(
map(val => {
if (val > 2) throw new Error("Invalid Value");
return val;
}),
retry(2)
)
.subscribe(
val => console.log(val),
err => console.log(err),
() => console.log("Complete")
);
}
}
Retry without any argument, will retry indefinitely
interval(1000)
.pipe(
map(val => {
if (val > 2) throw new Error("Invalid Value");
return val;
}),
retry()
)
.subscribe(val => console.log(val));
Retry(0) never retries.
interval(1000)
.pipe(
map(val => {
if (val > 2) throw new Error("Invalid Value");
return val;
}),
retry(0)
)
.subscribe(val => console.log(val));
The RetryWhen Angule RxJs operator retries the failed Observable every time a Notification Observable emits the next value.
Syntax
retryWhen<T>(notifier: (errors: Observable<any>) => Observable<any>): MonoTypeOperatorFunction<T>
Where notifier is the callback, which returns the Notification Observable
We register the notifier callback with the ReTryWhen Operator.
The notifier gets the errors observable as the argument, which emits whenever the source observable errors
We return the notifier observable, which we build using the errors observable.
The ReTryWhen subscribes to this notifier observable and how it behaves depends on the value emitted by the notifier observable
In the following example, map operator throws an error if the val > 2.
The errors are caught by retryWhen. It gets the error observable as its input. We use the pipe operator to add a tap to print the Retrying message on the console.
This code runs indefinitely as the source will always errors out and there is nothing to stop retryWhen
import { Component } from "@angular/core";
import { map, retryWhen, tap, take } from "rxjs/operators";
import { interval, Observable, observable, of, throwError, timer } from "rxjs";
@Component({
selector: "my-app",
template: `
<h1>Retry And ReTryWhen Example</h1>
`,
styleUrls: ["./app.component.css"]
})
export class AppComponent {
ngOnInit() {
interval(1000)
.pipe(
map(val => {
if (val > 2) throw new Error("Invalid Value");
return val;
}),
retryWhen(
error => error.pipe(
tap(() => console.log("Retrying... "))))
)
.subscribe(
val => console.log(val),
err => console.log(err),
() => console.log("Complete")
);
}
}
**Console **
0
1
2
Retrying..
0
1
2
Retrying..
Returning the error observable as it is also retries the source indefinitely
interval(1000)
.pipe(
map(val => {
if (val > 2) throw new Error("Invalid Value");
return val;
}),
retryWhen(error => error)
)
.subscribe(
val => console.log(val),
err => console.log(err),
() => console.log("Complete")
);
The following code adds delay of 2000 ms.
interval(1000)
.pipe(
map(val => {
if (val > 2) throw new Error("Invalid Value");
return val;
}),
retryWhen(
error =>
error.pipe(
tap(() => console.log("error occurred ")),
delay(2000),
tap(() => console.log("Retrying ..."))
)
)
)
.subscribe(
val => console.log(val),
err => console.log(err),
() => console.log("Complete")
);
As long as the notifier observable emits a value, the retryWhen will re subscribes to the source again.
In the following example, we switch observable to a new observable using SwitchMap operator. The new observable (of Operator) emits the value 1 after a delay of 1000 ms and the source is resubscribed again.
interval(1000)
.pipe(
map(val => {
if (val > 2) throw new Error("Invalid Value");
return val;
}),
retryWhen(error =>
error.pipe(
switchMap(() =>
of(1).pipe(
delay(1000),
tap(() => console.log("of emitted"))
)
),
tap(() => console.log("Retrying ..."))
)
)
)
.subscribe(
val => console.log(val),
err => console.log(err),
() => console.log("Complete")
);
The following uses of() instead of(1). of() does not emit any values but sends a complete notification on subscription. In this case, RetryWhen does not retry the source but completes. The subscribers will not receive any notification.
interval(1000)
.pipe(
map(val => {
if (val > 2) throw new Error("Error: Value greater than 2");
return val;
}),
retryWhen(error =>
error.pipe(
switchMap(() =>
of().pipe(
delay(1000),
tap(() => console.log("of emitted"))
)
),
tap(() => console.log("Retrying ..."))
)
)
)
.subscribe(
val => console.log(val),
err => console.error(err),
() => console.log("Complete")
);
Here is another interesting example where we switch to interval(3000) observable.
When the error occurs the first time in the source, RetryWhen triggers the interval(3000). When it emits a value source is re-subscribed.
But, the interval(3000) is going to emit another value after 3000ms. In this case, the RetryWhen will re subscribes to the source again, even if the source has errored out or already completed.
interval(1000)
.pipe(
map(val => {
if (val > 2) throw Error;
return val;
}),
retryWhen(error =>
error.pipe(
tap(() => console.log("error tapped")),
switchMap(() =>
interval(3000).pipe(tap(() => console.log("interval")))
),
tap(() => console.log("Retrying ..."))
)
)
)
.subscribe(
val => console.log(val),
err => console.log(err),
() => console.log("Complete")
);
*** Console ***
0
1
2
error tapped
interval
Retrying ...
0
1
interval
Retrying ...
0
1
interval
Retrying ...
In the example below, we use the scan operator to count the number of tries and throw an error if the number of attempts exceeds 2.
interval(1000)
.pipe(
map(val => {
if (val > 2) throw new Error("Invalid Value");
return val;
}),
retryWhen(error =>
error.pipe(
scan((acc, error) => {
if (acc > 2) throw error;
console.log("attempt " + acc);
return acc + 1;
}, 1),
delay(3000),
tap(() => console.log("Retrying ..."))
)
)
)
.subscribe(
val => console.log(val),
err => console.error(err),
() => console.log("Complete")
);
*** Console ***
0
1
2
attempt 1
Retrying ...
0
1
2
attempt 2
Retrying ...
0
1
2
Invalid Value
Use the dealyWhen to increase the time duration between each retries.
interval(1000)
.pipe(
map(val => {
if (val > 2) throw new Error("Invalid Value");
return val;
}),
retryWhen(error =>
error.pipe(
scan((acc, error) => {
if (acc > 2) throw error;
console.log("attempt " + acc);
return acc + 1;
}, 1),
delayWhen(val => timer(val * 2000)),
tap(() => console.log("Retrying ..."))
)
)
)
.subscribe(
val => console.log(val),
err => console.error(err),
() => console.log("Complete")
);
You can also use the take or takeWhile operator to stop the retries and emit the complete notification.
interval(1000)
.pipe(
map(val => {
if (val > 2) throw new Error("Invalid Value");
return val;
}),
retryWhen(error =>
error.pipe(
scan((acc, error) => {
console.log("attempt " + acc);
return acc + 1;
}, 1),
take(2),
delayWhen(val => timer(val * 2000)),
tap(() => console.log("Retrying ..."))
)
)
)
.subscribe(
val => console.log(val),
err => console.error(err),
() => console.log("Complete")
);
}
