Enum InferenceExecutionError 

#[non_exhaustive]
pub enum InferenceExecutionError {
    Provider(String),
    RateLimited {
        message: String,
        retry_after: Option<Duration>,
    },
    Overloaded {
        message: String,
        retry_after: Option<Duration>,
    },
    Timeout(String),
    StreamInterrupted {
        cause: InterruptCause,
        snapshot: Box<InterruptSnapshot>,
    },
    ContextOverflow(String),
    InvalidRequest(String),
    Unauthorized(String),
    ModelNotFound(String),
    ContentFiltered(String),
    AllModelsUnavailable,
    Cancelled,
}

Errors from LLM inference.

Variants split into three recoverability classes:

• Transient (retryable, count toward circuit breaker): RateLimited, Overloaded, Timeout, Provider, StreamInterrupted.
• Permanent (not retryable, do NOT count toward circuit breaker): ContextOverflow, InvalidRequest, Unauthorized, ModelNotFound, ContentFiltered.
• Fail-fast: AllModelsUnavailable, Cancelled.

Use InferenceExecutionError::is_retryable and InferenceExecutionError::counts_toward_circuit_breaker for policy decisions instead of pattern-matching variants directly where possible.

Variants (Non-exhaustive)

Non-exhaustive enums could have additional variants added in future. Therefore, when matching against variants of non-exhaustive enums, an extra wildcard arm must be added to account for any future variants.

Provider(String)

RateLimited

Fields

message: String

retry_after: Option<Duration>

Duration from the provider’s Retry-After header, if any.

Overloaded

Fields

message: String

retry_after: Option<Duration>

Timeout(String)

StreamInterrupted

Fields

cause: InterruptCause

snapshot: Box<InterruptSnapshot>

ContextOverflow(String)

InvalidRequest(String)

Unauthorized(String)

ModelNotFound(String)

ContentFiltered(String)

AllModelsUnavailable

Cancelled

Implementations

impl InferenceExecutionError

pub fn rate_limited(message: impl Into<String>) -> InferenceExecutionError

Short constructor for a rate-limit error with no Retry-After.

pub fn overloaded(message: impl Into<String>) -> InferenceExecutionError

Short constructor for an overloaded error with no Retry-After.

pub fn is_retryable(&self) -> bool

Whether the retry subsystem should try this request again.

Transient errors return true; permanent and fail-fast errors (including Cancelled) return false.

pub fn counts_toward_circuit_breaker(&self) -> bool

Whether this failure should increment the per-model circuit-breaker failure counter. Permanent errors (bad auth, bad schema, context overflow) must not trip the breaker — they would have failed with the same error on any model.

pub fn retry_after(&self) -> Option<Duration>

If this error carries a Retry-After hint from the provider, return it.

Trait Implementations

impl Clone for InferenceExecutionError

fn clone(&self) -> InferenceExecutionError

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for InferenceExecutionError

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

Formats the value using the given formatter.

impl Display for InferenceExecutionError

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

Formats the value using the given formatter.

impl Error for InferenceExecutionError

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.

impl From<InferenceExecutionError> for AgentLoopError

fn from(e: InferenceExecutionError) -> AgentLoopError

Converts to this type from the input type.