retry

Simple library for retry mechanism

Slightly inspired by Try::Tiny::Retry

SYNOPSIS

HTTP GET with retry:

url := "http://example.com"
var body []byte

err := retry.New(
    retry.Attempts(5),
    retry.Delay(100*time.Millisecond),
).Do(
    func() error {
        resp, err := http.Get(url)
        if err != nil {
            return err
        }
        defer resp.Body.Close()
        body, err = ioutil.ReadAll(resp.Body)
        if err != nil {
            return err
        }
        return nil
    },
)

if err != nil {
    // handle error
}

fmt.Println(string(body))

HTTP GET with retry with data:

url := "http://example.com"

body, err := retry.DoWithData(retry.New(),
    func() ([]byte, error) {
        resp, err := http.Get(url)
        if err != nil {
            return nil, err
        }
        defer resp.Body.Close()
        body, err := ioutil.ReadAll(resp.Body)
        if err != nil {
            return nil, err
        }

        return body, nil
    },
)

if err != nil {
    // handle error
}

fmt.Println(string(body))

Reusable retrier for high-frequency retry operations:

// Create retrier once, reuse many times
retrier := retry.New(
    retry.Attempts(5),
    retry.Delay(100*time.Millisecond),
)

// Minimal allocations in happy path
for {
    err := retrier.Do(
        func() error {
            return doWork()
        },
    )
    if err != nil {
        // handle error
    }
}

More examples

SEE ALSO

• codeGROOVE-dev/retry - Modern fork of avast/retry-go/v4 focused on correctness, reliability and efficiency. 100% API-compatible drop-in replacement. Looks really good.

• giantswarm/retry-go - slightly complicated interface.

• sethgrid/pester - only http retry for http calls with retries and backoff

• cenkalti/backoff - Go port of the exponential backoff algorithm from Google's HTTP Client Library for Java. Really complicated interface.

• rafaeljesus/retry-go - looks good, slightly similar as this package, don't have 'simple' Retry method

• matryer/try - very popular package, nonintuitive interface (for me)

BREAKING CHANGES

• 5.0.0

  • Complete API redesign: method-based retry operations
  • Renamed Config type to Retrier
  • Renamed NewConfig() to New()
  • Changed from package-level functions to methods: retry.Do(func, config) → retry.New(opts...).Do(func)
  • DelayTypeFunc signature changed: func(n uint, err error, config *Config) → func(n uint, err error, r *Retrier)
  • Migration: retry.Do(func, opts...) → retry.New(opts...).Do(func) (simple find & replace)
  • This change improves performance, simplifies the API, and provides a cleaner interface
  • Unwrap() now returns []error instead of error to support Go 1.20 multiple error wrapping.
  • errors.Unwrap(err) will now return nil (same as errors.Join). Use errors.Is or errors.As to inspect wrapped errors.

• 4.0.0

  • infinity retry is possible by set Attempts(0) by PR #49

• 3.0.0

  • DelayTypeFunc accepts a new parameter err - this breaking change affects only your custom Delay Functions. This change allow make delay functions based on error.

• 1.0.2 -> 2.0.0

  • argument of retry.Delay is final delay (no multiplication by retry.Units anymore)
  • function retry.Units are removed
  • more about this breaking change

• 0.3.0 -> 1.0.0

  • retry.Retry function are changed to retry.Do function
  • retry.RetryCustom (OnRetry) and retry.RetryCustomWithOpts functions are now implement via functions produces Options (aka retry.OnRetry)

Usage

func BackOffDelay

func BackOffDelay(n uint, _ error, config DelayContext) time.Duration

BackOffDelay is a DelayType which increases delay between consecutive retries

func FixedDelay

func FixedDelay(_ uint, _ error, config DelayContext) time.Duration

FixedDelay is a DelayType which keeps delay the same through all iterations

func FullJitterBackoffDelay

func FullJitterBackoffDelay(n uint, err error, config DelayContext) time.Duration

FullJitterBackoffDelay is a DelayTypeFunc that calculates delay using exponential backoff with full jitter. The delay is a random value between 0 and the current backoff ceiling. Formula: sleep = random_between(0, min(cap, base * 2^attempt)) It uses config.Delay as the base delay and config.MaxDelay as the cap.

func IsRecoverable

func IsRecoverable(err error) bool

IsRecoverable checks if error is an instance of unrecoverableError

func RandomDelay

func RandomDelay(_ uint, _ error, config DelayContext) time.Duration

RandomDelay is a DelayType which picks a random delay up to maxJitter

func Unrecoverable

func Unrecoverable(err error) error

Unrecoverable wraps an error in unrecoverableError struct

type DelayContext

type DelayContext interface {
    Delay() time.Duration
    MaxJitter() time.Duration
    MaxBackOffN() uint
    MaxDelay() time.Duration
}

DelayContext provides configuration values needed for delay calculation.

type DelayTypeFunc

type DelayTypeFunc func(n uint, err error, config DelayContext) time.Duration

DelayTypeFunc is called to return the next delay to wait after the retriable function fails on err after n attempts.

func CombineDelay

func CombineDelay(delays ...DelayTypeFunc) DelayTypeFunc

CombineDelay is a DelayType the combines all of the specified delays into a new DelayTypeFunc

type Error

type Error []error

Error type represents list of errors in retry

func (Error) As

func (e Error) As(target interface{}) bool

func (Error) Error

func (e Error) Error() string

Error method return string representation of Error It is an implementation of error interface

func (Error) Is

func (e Error) Is(target error) bool

func (Error) LastError

func (e Error) LastError() error

LastError returns the last error in the error list.

This is a convenience method for users migrating from retry-go v4.x where errors.Unwrap(err) returned the last error. In v5.0.0, errors.Unwrap(err) returns nil due to the switch to Unwrap() []error for Go 1.20 compatibility.

Migration example:

// v4.x code:
lastErr := errors.Unwrap(retryErr)

// v5.0.0 code (option 1 - recommended):
if errors.Is(retryErr, specificError) { ... }

// v5.0.0 code (option 2 - if you need the last error):
lastErr := retryErr.(retry.Error).LastError()

Note: Using errors.Is or errors.As is preferred as they check ALL wrapped errors, not just the last one.

func (Error) Unwrap

func (e Error) Unwrap() []error

Unwrap returns the list of errors that this Error is wrapping.

This method implements the Unwrap() []error interface introduced in Go 1.20 for multi-error unwrapping. This allows errors.Is and errors.As to traverse all wrapped errors, not just the last one.

IMPORTANT: errors.Unwrap(err) will return nil because the standard library's errors.Unwrap function only calls Unwrap() error, not Unwrap() []error. This is the same behavior as errors.Join in Go 1.20.

Example - Use errors.Is to check for specific errors:

err := retry.New(retry.Attempts(3)).Do(func() error {
    return os.ErrNotExist
})
if errors.Is(err, os.ErrNotExist) {
    // Handle not exist error
}

Example - Use errors.As to extract error details:

var pathErr *fs.PathError
if errors.As(err, &pathErr) {
    fmt.Println("Failed at path:", pathErr.Path)
}

Example - Get the last error directly (for migration):

if retryErr, ok := err.(retry.Error); ok {
    lastErr := retryErr.LastError()
}

See also: LastError() for direct access to the last error.

func (Error) WrappedErrors

func (e Error) WrappedErrors() []error

WrappedErrors returns the list of errors that this Error is wrapping. It is an implementation of the errwrap.Wrapper interface in package errwrap so that retry.Error can be used with that library.

type OnRetryFunc

type OnRetryFunc func(attempt uint, err error)

Function signature of OnRetry function

type Option

type Option func(*retrierCore)

Option represents an option for retry.

func Attempts

func Attempts(attempts uint) Option

Attempts set count of retry. Setting to 0 will retry until the retried function succeeds. default is 10

func AttemptsForError

func AttemptsForError(attempts uint, err error) Option

AttemptsForError sets count of retry in case execution results in given err Retries for the given err are also counted against total retries. The retry will stop if any of given retries is exhausted.

added in 4.3.0

func Context

func Context(ctx context.Context) Option

Context allow to set context of retry default are Background context

example of immediately cancellation (maybe it isn't the best example, but it describes behavior enough; I hope)

ctx, cancel := context.WithCancel(context.Background())
cancel()

retry.New().Do(
    func() error {
        ...
    },
    retry.Context(ctx),
)

func Delay

func Delay(delay time.Duration) Option

Delay set delay between retry default is 100ms

func DelayType

func DelayType(delayType DelayTypeFunc) Option

DelayType set type of the delay between retries default is a combination of BackOffDelay and RandomDelay for exponential backoff with jitter

func LastErrorOnly

func LastErrorOnly(lastErrorOnly bool) Option

return the direct last error that came from the retried function default is false (return wrapped errors with everything)

func MaxDelay

func MaxDelay(maxDelay time.Duration) Option

MaxDelay set maximum delay between retry does not apply by default

func MaxJitter

func MaxJitter(maxJitter time.Duration) Option

MaxJitter sets the maximum random Jitter between retries for RandomDelay

func OnRetry

func OnRetry(onRetry OnRetryFunc) Option

OnRetry function callback are called each retry

log each retry example:

retry.New().Do(
    func() error {
        return errors.New("some error")
    },
    retry.OnRetry(func(n uint, err error) {
        log.Printf("#%d: %s\n", n, err)
    }),
)

func RetryIf

func RetryIf(retryIf RetryIfFunc) Option

RetryIf controls whether a retry should be attempted after an error (assuming there are any retry attempts remaining)

skip retry if special error example:

retry.New().Do(
    func() error {
        return errors.New("special error")
    },
    retry.RetryIf(func(err error) bool {
        if err.Error() == "special error" {
            return false
        }
        return true
    })
)

By default RetryIf stops execution if the error is wrapped using retry.Unrecoverable, so above example may also be shortened to:

retry.New().Do(
    func() error {
        return retry.Unrecoverable(errors.New("special error"))
    }
)

func UntilSucceeded

func UntilSucceeded() Option

UntilSucceeded will retry until the retried function succeeds. Equivalent to setting Attempts(0).

func WithTimer

func WithTimer(t Timer) Option

WithTimer provides a way to swap out timer module implementations. This primarily is useful for mocking/testing, where you may not want to explicitly wait for a set duration for retries.

example of augmenting time.After with a print statement

type struct MyTimer {}

func (t *MyTimer) After(d time.Duration) <- chan time.Time {
    fmt.Print("Timer called!")
    return time.After(d)
}

retry.New().Do(
    func() error { ... },
       retry.WithTimer(&MyTimer{})
)

func WrapContextErrorWithLastError

func WrapContextErrorWithLastError(wrapContextErrorWithLastError bool) Option

WrapContextErrorWithLastError allows the context error to be returned wrapped with the last error that the retried function returned. This is only applicable when Attempts is set to 0 to retry indefinitly and when using a context to cancel / timeout

default is false

ctx, cancel := context.WithCancel(context.Background())
defer cancel()

retry.New().Do(
    func() error {
        ...
    },
    retry.Context(ctx),
    retry.Attempts(0),
    retry.WrapContextErrorWithLastError(true),
)

type Retrier

type Retrier struct {
}

Retrier is for retry operations that return only an e