gcp_sdk_gax::retry_policy

Trait RetryPolicy

Source 
pub trait RetryPolicy:
    Send
    + Sync
    + Debug {
    // Required method
    fn on_error(
        &self,
        loop_start: Instant,
        attempt_count: u32,
        idempotent: bool,
        error: Error,
    ) -> LoopState;

    // Provided methods
    fn on_throttle(
        &self,
        _loop_start: Instant,
        _attempt_count: u32,
    ) -> Option<Error> { ... }
    fn remaining_time(
        &self,
        _loop_start: Instant,
        _attempt_count: u32,
    ) -> Option<Duration> { ... }
}
Expand description

Determines how errors are handled in the retry loop.

Implementations of this trait determine if errors are retryable, and for how long the retry loop may continue.

Required Methods§

Source

fn on_error( &self, loop_start: Instant, attempt_count: u32, idempotent: bool, error: Error, ) -> LoopState

Query the retry policy after an error.

§Parameters
  • loop_start - when the retry loop started.
  • attempt_count - the number of attempts. This includes the initial attempt. This method called after the first attempt, so the value is always non-zero.
  • idempotent - if true assume the operation is idempotent. Many more errors are retryable on idempotent operations.
  • error - the last error when attempting the request.

Provided Methods§

Source

fn on_throttle( &self, _loop_start: Instant, _attempt_count: u32, ) -> Option<Error>

Query the retry policy after a retry attempt is throttled.

Retry attempts may be throttled before they are even sent out. The retry policy may choose to treat these as normal errors, consuming attempts, or may prefer to ignore them and always return LoopState::Continue.

§Parameters
  • loop_start - when the retry loop started.
  • attempt_count - the number of attempts. This method is never called before the first attempt.
Source

fn remaining_time( &self, _loop_start: Instant, _attempt_count: u32, ) -> Option<Duration>

The remaining time in the retry policy.

For policies based on time, this returns the remaining time in the policy. The retry loop can use this value to adjust the next RPC timeout. For policies that are not time based this returns None.

§Parameters
  • loop_start - when the retry loop started.
  • attempt_count - the number of attempts. This method is called before the first attempt, so the first value is zero.

Implementors§