Skip to content

fix(openai): cancel timed-out generation before releasing its lane #1150

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
i386 merged 5 commits into from
Aug 5, 2026
Merged

fix(openai): cancel timed-out generation before releasing its lane #1150

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
cff83a4
fix(openai): cancel timed-out generation before releasing its lane
i386 Aug 2, 2026
c759cf0
fix(skippy): admit fresh sessions before batch query
i386 Aug 2, 2026
9a3703f
fix(skippy): distinguish session batch-size lookups
i386 Aug 4, 2026
b536993
fix(skippy): accept mutable active batch lookup
ndizazzo Aug 4, 2026
b95c08d
fix(skippy): stop cancelled generation before token output
i386 Aug 5, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
41 changes: 38 additions & 3 deletions crates/openai-frontend/src/backend.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,6 +8,7 @@ use std::{

use async_trait::async_trait;
use futures_core::Stream;
use tokio::sync::Notify;

use crate::{
chat::{ChatCompletionChunk, ChatCompletionRequest, ChatCompletionResponse},
Expand All @@ -23,9 +24,15 @@ pub type CompletionStream =

pub type OpenAiResult<T> = Result<T, OpenAiError>;

#[derive(Debug, Default)]
struct CancellationState {
cancelled: AtomicBool,
notify: Notify,
}

#[derive(Debug, Clone, Default)]
pub struct CancellationToken {
cancelled: Arc<AtomicBool>,
state: Arc<CancellationState>,
}

impl CancellationToken {
Expand All @@ -34,11 +41,23 @@ impl CancellationToken {
}

pub fn cancel(&self) {
self.cancelled.store(true, Ordering::Relaxed);
if !self.state.cancelled.swap(true, Ordering::AcqRel) {
self.state.notify.notify_waiters();
}
}

pub fn is_cancelled(&self) -> bool {
self.cancelled.load(Ordering::Relaxed)
self.state.cancelled.load(Ordering::Acquire)
}

pub async fn cancelled(&self) {
loop {
let notified = self.state.notify.notified();
if self.is_cancelled() {
return;
}
notified.await;
}
}
}

Expand Down Expand Up @@ -74,6 +93,14 @@ pub trait OpenAiBackend: Send + Sync + 'static {
request: ChatCompletionRequest,
) -> OpenAiResult<ChatCompletionResponse>;

async fn chat_completion_with_context(
&self,
request: ChatCompletionRequest,
_context: OpenAiRequestContext,
) -> OpenAiResult<ChatCompletionResponse> {
self.chat_completion(request).await
}

async fn chat_completion_stream(
&self,
request: ChatCompletionRequest,
Expand All @@ -86,6 +113,14 @@ pub trait OpenAiBackend: Send + Sync + 'static {
))
}

async fn completion_with_context(
&self,
request: CompletionRequest,
_context: OpenAiRequestContext,
) -> OpenAiResult<CompletionResponse> {
self.completion(request).await
}

async fn completion_stream(
&self,
_request: CompletionRequest,
Expand Down
22 changes: 20 additions & 2 deletions crates/openai-frontend/src/guardrails/compact.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -63,9 +63,18 @@ impl OpenAiBackend for CompactingOpenAiBackend {
async fn chat_completion(
&self,
request: ChatCompletionRequest,
) -> OpenAiResult<ChatCompletionResponse> {
self.chat_completion_with_context(request, OpenAiRequestContext::new())
.await
}

async fn chat_completion_with_context(
&self,
request: ChatCompletionRequest,
context: OpenAiRequestContext,
) -> OpenAiResult<ChatCompletionResponse> {
self.backend
.chat_completion(self.compact_request(request)?)
.chat_completion_with_context(self.compact_request(request)?, context)
.await
}

Expand All @@ -80,7 +89,16 @@ impl OpenAiBackend for CompactingOpenAiBackend {
}

async fn completion(&self, request: CompletionRequest) -> OpenAiResult<CompletionResponse> {
self.backend.completion(request).await
self.completion_with_context(request, OpenAiRequestContext::new())
.await
}

async fn completion_with_context(
&self,
request: CompletionRequest,
context: OpenAiRequestContext,
) -> OpenAiResult<CompletionResponse> {
self.backend.completion_with_context(request, context).await
}

async fn completion_stream(
Expand Down
37 changes: 31 additions & 6 deletions crates/openai-frontend/src/guardrails/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,6 +75,7 @@ impl GuardedOpenAiBackend {
async fn guarded_chat_completion(
&self,
request: ChatCompletionRequest,
context: OpenAiRequestContext,
) -> OpenAiResult<ChatCompletionResponse> {
let _guardrail_error_catalog = guardrail_error_catalog();
let policy = self.policy.snapshot();
Expand All @@ -90,13 +91,15 @@ impl GuardedOpenAiBackend {
GuardrailTelemetryOutcome::PassThrough,
None,
);
self.backend.chat_completion(request).await
self.backend
.chat_completion_with_context(request, context)
.await
}
GuardrailRequestOutcome::Reject { kind } => Err(errors::guardrail_error(*kind)),
GuardrailRequestOutcome::Guarded { backend_request } => {
if matches!(policy.mode, GuardrailMode::MetricsOnly) {
return self
.metrics_only_chat_completion(request, &engine, &prepared)
.metrics_only_chat_completion(request, &engine, &prepared, context)
.await;
}

Expand All @@ -107,7 +110,7 @@ impl GuardedOpenAiBackend {
loop {
let response = self
.backend
.chat_completion(attempt_request.clone())
.chat_completion_with_context(attempt_request.clone(), context.clone())
.await?;
let classified = engine.classify_response(&prepared, &response);
let contract = telemetry_contract(&prepared.state.request_contract);
Expand Down Expand Up @@ -165,8 +168,12 @@ impl GuardedOpenAiBackend {
request: ChatCompletionRequest,
engine: &GuardrailEngine,
prepared: &state::PreparedGuardrailRequest,
context: OpenAiRequestContext,
) -> OpenAiResult<ChatCompletionResponse> {
let response = self.backend.chat_completion(request).await?;
let response = self
.backend
.chat_completion_with_context(request, context)
.await?;
let classified = engine.classify_response(prepared, &response);
self.record_outcome(
prepared.state.mode,
Expand Down Expand Up @@ -303,7 +310,16 @@ impl OpenAiBackend for GuardedOpenAiBackend {
&self,
request: ChatCompletionRequest,
) -> OpenAiResult<ChatCompletionResponse> {
self.guarded_chat_completion(request).await
self.guarded_chat_completion(request, OpenAiRequestContext::new())
.await
}

async fn chat_completion_with_context(
&self,
request: ChatCompletionRequest,
context: OpenAiRequestContext,
) -> OpenAiResult<ChatCompletionResponse> {
self.guarded_chat_completion(request, context).await
}

async fn chat_completion_stream(
Expand All @@ -315,7 +331,16 @@ impl OpenAiBackend for GuardedOpenAiBackend {
}

async fn completion(&self, request: CompletionRequest) -> OpenAiResult<CompletionResponse> {
self.backend.completion(request).await
self.completion_with_context(request, OpenAiRequestContext::new())
.await
}

async fn completion_with_context(
&self,
request: CompletionRequest,
context: OpenAiRequestContext,
) -> OpenAiResult<CompletionResponse> {
self.backend.completion_with_context(request, context).await
}

async fn completion_stream(
Expand Down
24 changes: 22 additions & 2 deletions crates/openai-frontend/src/hooks.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -128,12 +128,23 @@ impl OpenAiBackend for HookedOpenAiBackend {
}

async fn chat_completion(
&self,
request: ChatCompletionRequest,
) -> OpenAiResult<ChatCompletionResponse> {
self.chat_completion_with_context(request, OpenAiRequestContext::new())
.await
}

async fn chat_completion_with_context(
&self,
mut request: ChatCompletionRequest,
context: OpenAiRequestContext,
) -> OpenAiResult<ChatCompletionResponse> {
let outcome = self.hooks.before_chat_completion(&mut request).await?;
apply_chat_hook_outcome(&mut request, &outcome);
self.backend.chat_completion(request).await
self.backend
.chat_completion_with_context(request, context)
.await
}

async fn chat_completion_stream(
Expand All @@ -147,7 +158,16 @@ impl OpenAiBackend for HookedOpenAiBackend {
}

async fn completion(&self, request: CompletionRequest) -> OpenAiResult<CompletionResponse> {
self.backend.completion(request).await
self.completion_with_context(request, OpenAiRequestContext::new())
.await
}

async fn completion_with_context(
&self,
request: CompletionRequest,
context: OpenAiRequestContext,
) -> OpenAiResult<CompletionResponse> {
self.backend.completion_with_context(request, context).await
}

async fn completion_stream(
Expand Down
Loading
Loading