Last retry for a request sleeps unnecessarily

[nojnhuh]

Once a request is found to have failed and to be retried, the applicable HTTP 429 Retry-After or exponential backoff logic kicks in and sleeps for the appropriate amount of time before trying the request again. This sleep occurs even after the last attempt which only delays callers from receiving the ultimate response. Any sleeping should only occur when there is at least one retry attempt remaining.

ref:

go-autorest/autorest/sender.go

Line 319 in 2fa44cb

func doRetryForStatusCodesImpl(s Sender, r *http.Request, count429 bool, attempts int, backoff, cap time.Duration, codes ...int) (resp *http.Response, err error) {

Combined with #712, this causes any request consistently failing with a retry-able status code to take 90s+ with the default 30s RetryDuration on clients (30s for the first request + 60s for the second request). Made only once and without sleeping after, I see the same request take ~3s.

[jhendrixMSFT]

Closing due to staleness. In addition, these modules are no longer supported. See https://github.com/Azure/go-autorest?tab=readme-ov-file#migrating-to-supported-sdks for more info.

This has been fixed in the new Azure SDKs that are based on azcore.