Enum OrchestratorError 

pub enum OrchestratorError {
    ChannelClosed,
    Inference(String),
    ConfigError(String),
    RateLimited {
        retry_after_secs: u64,
    },
    BudgetExceeded {
        spent: f64,
        limit: f64,
    },
    AuthFailed(String),
    InferenceTimeout {
        timeout_secs: u64,
    },
    Other(String),
}

Orchestrator-specific errors.

All variants are non-panicking. Callers should match on the variant to decide whether to retry, shed, or propagate the error.

Variants

ChannelClosed

A pipeline channel was closed before the request could be delivered.

This typically means a pipeline stage task has exited. The pipeline should be restarted. This error is not retryable within the same pipeline instance.

Inference(String)

An inference worker returned an error.

The inner string contains the provider error message. May be retryable depending on the underlying cause (transient network vs. invalid request).

ConfigError(String)

Pipeline or worker configuration is invalid.

Returned during startup validation. Not retryable without a config change.

RateLimited

Provider returned HTTP 429 — callers should back off for retry_after.

Fields

retry_after_secs: u64

Seconds to wait before retrying, parsed from Retry-After header.

BudgetExceeded

Spending cap reached — no further inference allowed this session.

Fields

spent: f64

Amount spent so far in USD.

limit: f64

Configured spending limit in USD.

AuthFailed(String)

Provider rejected the request due to an authentication failure.

Returned when the provider returns HTTP 401 or 403. Not retryable without a credential rotation.

InferenceTimeout

The inference call exceeded the configured timeout and was cancelled.

The timeout_secs field reflects the timeout that was breached. Retryable with a longer timeout or by shedding the request.

Fields

timeout_secs: u64

Configured timeout that was exceeded.

Other(String)

A catch-all error variant for errors that do not fit the other categories.

Implementations

impl OrchestratorError

pub fn error_kind(&self) -> &'static str

Return a stable lowercase string identifying the error variant, suitable for use as a metric label or structured log field.

Panics

This function does not panic.

pub fn is_retryable(&self) -> bool

Return true if retrying the request may succeed.

Transient errors (RateLimited, InferenceTimeout, Inference) are retryable. Permanent errors (ConfigError, AuthFailed, BudgetExceeded, ChannelClosed) are not.

Panics

This function does not panic.

Examples

use tokio_prompt_orchestrator::OrchestratorError;

assert!(OrchestratorError::RateLimited { retry_after_secs: 5 }.is_retryable());
assert!(!OrchestratorError::AuthFailed("bad key".into()).is_retryable());

Trait Implementations

impl Debug for OrchestratorError

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Display for OrchestratorError

fn fmt(&self, __formatter: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Error for OrchestratorError

fn source(&self) -> Option<&(dyn Error + 'static)>

Returns the lower-level source of this error, if any.

fn description(&self) -> &str

👎Deprecated since 1.42.0:

use the Display impl or to_string()

fn cause(&self) -> Option<&dyn Error>

👎Deprecated since 1.33.0:

replaced by Error::source, which can support downcasting

fn provide<'a>(&'a self, request: &mut Request<'a>)

🔬This is a nightly-only experimental API. (error_generic_member_access)

Provides type-based access to context intended for error reports.