agent/codex
2
0
Fork 0
mirror of https://github.com/pchuan98/codex.git synced 2026-07-01 00:31:56 +08:00
Code Issues Packages Projects Releases Wiki Activity

[rollout-trace] Include x-request-id in rollout trace. (#20066)

Browse Source 
## Why

Rollout traces need an identifier that can be used to correlate a Codex
inference with upstream Responses API, proxy, and engine logs. The
reduced trace model already exposed `upstream_request_id`, but it was
being populated from the Responses API `response.id`. That value is
useful for `previous_response_id` chaining, but it is not the transport
request id that upstream systems key on.

This PR separates those concepts so trace consumers can reliably answer
both questions:

- which Responses API response did this inference produce?
- which upstream request handled it?

## Structure

The change keeps the upstream request id at the same lifecycle level as
the provider stream:

- `codex-api` captures the `x-request-id` HTTP response header when the
SSE stream is created and exposes it on `ResponseStream`. Fixture and
websocket streams set the field to `None` because they do not have that
HTTP response header.
- `codex-core` carries that stream-level id into `InferenceTraceAttempt`
when recording terminal stream outcomes. Completed, failed, cancelled,
dropped-stream, and pre-response error paths all record the id when it
is available.
- `rollout-trace` now records both identifiers in raw terminal inference
events and response payloads: `response_id` for the Responses API
`response.id`, and `upstream_request_id` for `x-request-id`.
- The reducer stores both fields on `InferenceCall`. It also uses
`response_id` for `previous_response_id` conversation linking, which
removes the old accidental dependency on the misnamed
`upstream_request_id` field.
- Terminal inference reduction now consumes the full terminal payload
(`InferenceCompleted`, `InferenceFailed`, or `InferenceCancelled`) in
one place. That keeps status, partial payloads, response ids, and
upstream request ids consistent across success, failure, cancellation,
and late stream-mapper events.

## Why This Shape

`x-request-id` is a property of the HTTP/provider response envelope, not
an SSE event. Capturing it once in `codex-api` and plumbing it through
terminal trace recording avoids trying to infer the value from stream
contents, and it preserves the id even when the stream fails or is
cancelled after only partial output.

Keeping `response_id` separate from `upstream_request_id` also makes the
reduced trace model less surprising: `response_id` remains the
conversation-continuation id, while `upstream_request_id` is the
operational correlation id for upstream debugging.

## Validation

The PR updates trace and reducer coverage for:

- reading `x-request-id` from SSE response headers;
- storing the true upstream request id on completed inference calls;
- preserving upstream request ids for cancelled and late-cancelled
inference streams;
- keeping `previous_response_id` reconstruction tied to `response_id`
rather than transport request ids.
This commit is contained in:
cassirer-openai
2026-04-28 14:11:17 -07:00
committed by GitHub
Unverified
parent 10e2a73b3c
commit 89698ad1c3
17 changed files with 240 additions and 90 deletions
Download Patch File Download Diff File Expand all files Collapse all files
codex-rs/codex-api/src/common.rs
+2
View File
@@ -287,6 +287,8 @@ pub fn create_text_param_for_request(
pub struct ResponseStream {
pub rx_event: mpsc::Receiver<Result<ResponseEvent, ApiError>>,
/// Server-assigned `x-request-id` response header, when present.
pub upstream_request_id: Option<String>,
}
impl Stream for ResponseStream {
codex-rs/codex-api/src/endpoint/responses_websocket.rs
+4 -1
View File
@@ -279,7 +279,10 @@ impl ResponsesWebsocketConnection {
.instrument(current_span),
);
Ok(ResponseStream { rx_event })
Ok(ResponseStream {
rx_event,
upstream_request_id: None,
})
}
}
codex-rs/codex-api/src/sse/responses.rs
+17 -4
View File
@@ -28,6 +28,7 @@ use tracing::trace;
const X_REASONING_INCLUDED_HEADER: &str = "x-reasoning-included";
const OPENAI_MODEL_HEADER: &str = "openai-model";
const REQUEST_ID_HEADER: &str = "x-request-id";
const TRUSTED_ACCESS_FOR_CYBER_VERIFICATION: &str = "trusted_access_for_cyber";
/// Streams SSE events from an on-disk fixture for tests.
@@ -53,7 +54,10 @@ pub fn stream_from_fixture(
idle_timeout,
/*telemetry*/ None,
));
Ok(ResponseStream { rx_event })
Ok(ResponseStream {
rx_event,
upstream_request_id: None,
})
}
pub fn spawn_response_stream(
@@ -77,6 +81,11 @@ pub fn spawn_response_stream(
.headers
.get(X_REASONING_INCLUDED_HEADER)
.is_some();
let upstream_request_id = stream_response
.headers
.get(REQUEST_ID_HEADER)
.and_then(|value| value.to_str().ok())
.map(str::to_string);
if let Some(turn_state) = turn_state.as_ref()
&& let Some(header_value) = stream_response
.headers
@@ -104,7 +113,10 @@ pub fn spawn_response_stream(
process_sse(stream_response.bytes, tx_event, idle_timeout, telemetry).await;
});
ResponseStream { rx_event }
ResponseStream {
rx_event,
upstream_request_id,
}
}
#[derive(Debug, Deserialize)]
@@ -1058,8 +1070,9 @@ mod tests {
}
#[tokio::test]
async fn spawn_response_stream_emits_server_model_header() {
async fn spawn_response_stream_emits_header_events() {
let mut headers = HeaderMap::new();
headers.insert(REQUEST_ID_HEADER, HeaderValue::from_static("req-1"));
headers.insert(
OPENAI_MODEL_HEADER,
HeaderValue::from_static(CYBER_RESTRICTED_MODEL_FOR_TESTS),
@@ -1077,13 +1090,13 @@ mod tests {
/*telemetry*/ None,
/*turn_state*/ None,
);
assert_eq!(stream.upstream_request_id.as_deref(), Some("req-1"));
let event = stream
.rx_event
.recv()
.await
.expect("expected server model event")
.expect("expected ok event");
match event {
ResponseEvent::ServerModel(model) => {
assert_eq!(model, CYBER_RESTRICTED_MODEL_FOR_TESTS);
codex-rs/core/src/client.rs
+75 -13
View File
@@ -1234,8 +1234,13 @@ impl ModelClientSession {
Err(ApiError::Transport(
unauthorized_transport @ TransportError::Http { status, .. },
)) if status == StatusCode::UNAUTHORIZED => {
inference_trace_attempt
.record_failed(&unauthorized_transport, /*output_items*/ &[]);
let response_debug_context =
extract_response_debug_context(&unauthorized_transport);
inference_trace_attempt.record_failed(
&unauthorized_transport,
response_debug_context.request_id.as_deref(),
/*output_items*/ &[],
);
pending_retry = PendingUnauthorizedRetry::from_recovery(
handle_unauthorized(
unauthorized_transport,
@@ -1247,8 +1252,14 @@ impl ModelClientSession {
continue;
}
Err(err) => {
let response_debug_context =
extract_response_debug_context_from_api_error(&err);
let err = map_api_error(err);
inference_trace_attempt.record_failed(&err, /*output_items*/ &[]);
inference_trace_attempt.record_failed(
&err,
response_debug_context.request_id.as_deref(),
/*output_items*/ &[],
);
return Err(err);
}
}
@@ -1374,8 +1385,14 @@ impl ModelClientSession {
.stream_request(ws_request, self.websocket_session.connection_reused())
.await
.map_err(|err| {
let response_debug_context =
extract_response_debug_context_from_api_error(&err);
let err = map_api_error(err);
inference_trace_attempt.record_failed(&err, /*output_items*/ &[]);
inference_trace_attempt.record_failed(
&err,
response_debug_context.request_id.as_deref(),
/*output_items*/ &[],
);
err
})?;
let (stream, last_request_rx) = map_response_stream(
@@ -1636,7 +1653,29 @@ fn parent_thread_id_header_value(session_source: &SessionSource) -> Option<Strin
const RESPONSE_STREAM_CHANNEL_CAPACITY: usize = 1600;
const STREAM_DROPPED_REASON: &str = "response stream dropped before provider terminal event";
fn map_response_stream<S>(
fn map_response_stream(
api_stream: codex_api::ResponseStream,
session_telemetry: SessionTelemetry,
inference_trace_attempt: InferenceTraceAttempt,
) -> (ResponseStream, oneshot::Receiver<LastResponse>) {
let codex_api::ResponseStream {
rx_event,
upstream_request_id,
} = api_stream;
let api_stream = codex_api::ResponseStream {
rx_event,
upstream_request_id: None,
};
map_response_events(
upstream_request_id,
api_stream,
session_telemetry,
inference_trace_attempt,
)
}
fn map_response_events<S>(
upstream_request_id: Option<String>,
api_stream: S,
session_telemetry: SessionTelemetry,
inference_trace_attempt: InferenceTraceAttempt,
@@ -1658,10 +1697,15 @@ where
let mut tx_last_response = Some(tx_last_response);
let mut items_added: Vec<ResponseItem> = Vec::new();
let mut api_stream = api_stream;
let upstream_request_id = upstream_request_id.as_deref();
loop {
let event = tokio::select! {
_ = consumer_dropped.cancelled() => {
inference_trace_attempt.record_cancelled(STREAM_DROPPED_REASON, &items_added);
inference_trace_attempt.record_cancelled(
STREAM_DROPPED_REASON,
upstream_request_id,
&items_added,
);
return;
}
event = api_stream.next() => event,
@@ -1677,8 +1721,11 @@ where
.await
.is_err()
{
inference_trace_attempt
.record_cancelled(STREAM_DROPPED_REASON, &items_added);
inference_trace_attempt.record_cancelled(
STREAM_DROPPED_REASON,
upstream_request_id,
&items_added,
);
return;
}
}
@@ -1698,6 +1745,7 @@ where
}
inference_trace_attempt.record_completed(
&response_id,
upstream_request_id,
&token_usage,
&items_added,
);
@@ -1721,14 +1769,25 @@ where
}
Ok(event) => {
if tx_event.send(Ok(event)).await.is_err() {
inference_trace_attempt
.record_cancelled(STREAM_DROPPED_REASON, &items_added);
inference_trace_attempt.record_cancelled(
STREAM_DROPPED_REASON,
upstream_request_id,
&items_added,
);
return;
}
}
Err(err) => {
let response_debug_context =
extract_response_debug_context_from_api_error(&err);
let upstream_request_id =
upstream_request_id.or(response_debug_context.request_id.as_deref());
let mapped = map_api_error(err);
inference_trace_attempt.record_failed(&mapped, &items_added);
inference_trace_attempt.record_failed(
&mapped,
upstream_request_id,
&items_added,
);
if !logged_error {
session_telemetry.see_event_completed_failed(&mapped);
logged_error = true;
@@ -1739,8 +1798,11 @@ where
}
}
}
inference_trace_attempt
.record_failed("stream closed before response.completed", &items_added);
inference_trace_attempt.record_failed(
"stream closed before response.completed",
upstream_request_id,
&items_added,
);
});
(
codex-rs/core/src/client_tests.rs
+12 -2
View File
@@ -282,7 +282,12 @@ async fn dropped_response_stream_traces_cancelled_partial_output() -> anyhow::Re
let item = output_message("msg-1", "partial answer");
let api_stream = futures::stream::iter([Ok(ResponseEvent::OutputItemDone(item))])
.chain(futures::stream::pending());
let (mut stream, _) = super::map_response_stream(api_stream, test_session_telemetry(), attempt);
let (mut stream, _) = super::map_response_events(
/*upstream_request_id*/ None,
api_stream,
test_session_telemetry(),
attempt,
);
let observed = stream
.next()
@@ -332,7 +337,12 @@ async fn dropped_backpressured_response_stream_traces_cancelled_partial_output()
notify: Arc::clone(&backpressured_item_yielded),
};
let (stream, _) = super::map_response_stream(api_stream, test_session_telemetry(), attempt);
let (stream, _) = super::map_response_events(
/*upstream_request_id*/ None,
api_stream,
test_session_telemetry(),
attempt,
);
// Fill the mapper channel with non-terminal events, then yield one output
// item. The mapper has observed that item and is blocked trying to send it
codex-rs/rollout-trace/src/inference.rs
+24 -4
View File
@@ -84,6 +84,7 @@ struct EnabledInferenceTraceAttempt {
#[derive(Serialize)]
struct TracedResponseStreamOutput<'a> {
response_id: Option<&'a str>,
upstream_request_id: Option<&'a str>,
token_usage: Option<&'a TokenUsage>,
output_items: Vec<JsonValue>,
}
@@ -174,6 +175,7 @@ impl InferenceTraceAttempt {
pub fn record_completed(
&self,
response_id: &str,
upstream_request_id: Option<&str>,
token_usage: &Option<TokenUsage>,
output_items: &[ResponseItem],
) {
@@ -183,6 +185,7 @@ impl InferenceTraceAttempt {
let Some(response_payload) = write_response_payload_best_effort(
attempt,
Some(response_id),
upstream_request_id,
token_usage.as_ref(),
output_items,
) else {
@@ -194,13 +197,19 @@ impl InferenceTraceAttempt {
RawTraceEventPayload::InferenceCompleted {
inference_call_id: attempt.inference_call_id.clone(),
response_id: Some(response_id.to_string()),
upstream_request_id: upstream_request_id.map(str::to_string),
response_payload,
},
);
}
/// Records pre-response and mid-stream failures.
pub fn record_failed(&self, error: impl Display, output_items: &[ResponseItem]) {
pub fn record_failed(
&self,
error: impl Display,
upstream_request_id: Option<&str>,
output_items: &[ResponseItem],
) {
let Some(attempt) = self.take_terminal_attempt() else {
return;
};
@@ -210,6 +219,7 @@ impl InferenceTraceAttempt {
write_response_payload_best_effort(
attempt,
/*response_id*/ None,
upstream_request_id,
/*token_usage*/ None,
output_items,
)
@@ -218,6 +228,7 @@ impl InferenceTraceAttempt {
&attempt.context,
RawTraceEventPayload::InferenceFailed {
inference_call_id: attempt.inference_call_id.clone(),
upstream_request_id: upstream_request_id.map(str::to_string),
error: error.to_string(),
partial_response_payload,
},
@@ -229,7 +240,12 @@ impl InferenceTraceAttempt {
/// This happens when the turn is interrupted or when mailbox delivery
/// preempts the current sampling request. Complete output items observed
/// before that point are retained as partial response evidence.
pub fn record_cancelled(&self, reason: impl Display, output_items: &[ResponseItem]) {
pub fn record_cancelled(
&self,
reason: impl Display,
upstream_request_id: Option<&str>,
output_items: &[ResponseItem],
) {
let Some(attempt) = self.take_terminal_attempt() else {
return;
};
@@ -239,6 +255,7 @@ impl InferenceTraceAttempt {
write_response_payload_best_effort(
attempt,
/*response_id*/ None,
upstream_request_id,
/*token_usage*/ None,
output_items,
)
@@ -247,6 +264,7 @@ impl InferenceTraceAttempt {
&attempt.context,
RawTraceEventPayload::InferenceCancelled {
inference_call_id: attempt.inference_call_id.clone(),
upstream_request_id: upstream_request_id.map(str::to_string),
reason: reason.to_string(),
partial_response_payload,
},
@@ -312,11 +330,13 @@ fn write_json_payload_best_effort(
fn write_response_payload_best_effort(
attempt: &EnabledInferenceTraceAttempt,
response_id: Option<&str>,
upstream_request_id: Option<&str>,
token_usage: Option<&TokenUsage>,
output_items: &[ResponseItem],
) -> Option<crate::RawPayloadRef> {
let response_payload = TracedResponseStreamOutput {
response_id,
upstream_request_id,
token_usage,
output_items: output_items.iter().map(trace_response_item_json).collect(),
};
@@ -387,7 +407,7 @@ mod tests {
"content": [{"type": "input_text", "text": "hello"}]
}],
}));
attempt.record_completed("resp-1", &None, &[]);
attempt.record_completed("resp-1", Some("req-1"), &None, &[]);
let rollout = replay_bundle(temp.path())?;
let inference = rollout
@@ -400,7 +420,7 @@ mod tests {
assert_eq!(inference.thread_id, "thread-root");
assert_eq!(inference.codex_turn_id, "turn-1");
assert_eq!(inference.execution.status, ExecutionStatus::Completed);
assert_eq!(inference.upstream_request_id, Some("resp-1".to_string()));
assert_eq!(inference.upstream_request_id, Some("req-1".to_string()));
assert_eq!(rollout.raw_payloads.len(), 2);
Ok(())
codex-rs/rollout-trace/src/model/conversation.rs
+3 -1
View File
@@ -152,7 +152,9 @@ pub struct InferenceCall {
pub execution: ExecutionWindow,
pub model: String,
pub provider_name: String,
/// Upstream request ID returned by HTTP/proxy/engine infrastructure.
/// Responses API response id, used by follow-up `previous_response_id` requests.
pub response_id: Option<String>,
/// Request id returned by HTTP/proxy/engine infrastructure.
pub upstream_request_id: Option<String>,
/// Complete ordered input snapshot sent with this request.
pub request_item_ids: Vec<ConversationItemId>,
codex-rs/rollout-trace/src/raw_event.rs
+9
View File
@@ -101,17 +101,26 @@ pub enum RawTraceEventPayload {
},
InferenceCompleted {
inference_call_id: InferenceCallId,
/// Responses API `response.id`; used by `previous_response_id`.
response_id: Option<String>,
/// Provider transport request id, such as `x-request-id`.
upstream_request_id: Option<String>,
response_payload: RawPayloadRef,
},
InferenceFailed {
inference_call_id: InferenceCallId,
/// Provider transport request id, such as `x-request-id`, when the
/// provider returned one before the stream failed.
upstream_request_id: Option<String>,
error: String,
/// Partial response payload, when stream events arrived before failure.
partial_response_payload: Option<RawPayloadRef>,
},
InferenceCancelled {
inference_call_id: InferenceCallId,
/// Provider transport request id, such as `x-request-id`, when observed
/// before Codex stopped consuming the stream.
upstream_request_id: Option<String>,
/// Why Codex stopped consuming the provider stream before a terminal response event.
reason: String,
/// Completed output items observed before cancellation, if any.
codex-rs/rollout-trace/src/reducer/code_cell_tests.rs
+4
View File
@@ -64,6 +64,7 @@ fn code_cell_lifecycle_links_nested_tools_waits_and_outputs() -> anyhow::Result<
writer.append(RawTraceEventPayload::InferenceCompleted {
inference_call_id: "inference-1".to_string(),
response_id: Some("resp-1".to_string()),
upstream_request_id: None,
response_payload: response,
})?;
writer.append_with_context(
@@ -247,6 +248,7 @@ fn fast_code_cell_lifecycle_waits_for_source_item() -> anyhow::Result<()> {
writer.append(RawTraceEventPayload::InferenceCompleted {
inference_call_id: "inference-1".to_string(),
response_id: Some("resp-1".to_string()),
upstream_request_id: None,
response_payload: response,
})?;
@@ -301,6 +303,7 @@ fn cancelled_turn_terminates_unfinished_code_cell() -> anyhow::Result<()> {
writer.append(RawTraceEventPayload::InferenceCompleted {
inference_call_id: "inference-1".to_string(),
response_id: Some("resp-1".to_string()),
upstream_request_id: None,
response_payload: response,
})?;
writer.append_with_context(
@@ -388,6 +391,7 @@ fn runtime_code_cell_ids_can_repeat_across_threads() -> anyhow::Result<()> {
writer.append(RawTraceEventPayload::InferenceCompleted {
inference_call_id: inference_call_id.to_string(),
response_id: Some(format!("resp-{thread_id}")),
upstream_request_id: None,
response_payload: response,
})?;
writer.append_with_context(
codex-rs/rollout-trace/src/reducer/conversation.rs
+1 -1
View File
@@ -76,7 +76,7 @@ impl TraceReducer {
.values()
.find(|inference| {
inference.thread_id == thread_id
&& inference.upstream_request_id.as_deref() == Some(previous_response_id)
&& inference.response_id.as_deref() == Some(previous_response_id)
})
.map(|inference| {
let mut ids = inference.request_item_ids.clone();
codex-rs/rollout-trace/src/reducer/inference.rs
+69 -21
View File
@@ -13,6 +13,7 @@ use crate::model::InferenceCall;
use crate::model::InferenceCallId;
use crate::payload::RawPayloadRef;
use crate::raw_event::RawEventSeq;
use crate::raw_event::RawTraceEventPayload;
/// Raw inference-start fields after dispatch has stripped the common event envelope.
///
@@ -91,6 +92,7 @@ impl TraceReducer {
},
model: started.model,
provider_name: started.provider_name,
response_id: None,
upstream_request_id: None,
request_item_ids,
response_item_ids: Vec::new(),
@@ -137,11 +139,49 @@ impl TraceReducer {
&mut self,
seq: RawEventSeq,
wall_time_unix_ms: i64,
inference_call_id: InferenceCallId,
status: ExecutionStatus,
response_id: Option<String>,
response_payload: Option<RawPayloadRef>,
payload: RawTraceEventPayload,
) -> Result<()> {
let (inference_call_id, status, response_id, upstream_request_id, response_payload) =
match payload {
RawTraceEventPayload::InferenceCompleted {
inference_call_id,
response_id,
upstream_request_id,
response_payload,
} => (
inference_call_id,
ExecutionStatus::Completed,
response_id,
upstream_request_id,
Some(response_payload),
),
RawTraceEventPayload::InferenceFailed {
inference_call_id,
upstream_request_id,
partial_response_payload,
..
} => (
inference_call_id,
ExecutionStatus::Failed,
None,
upstream_request_id,
partial_response_payload,
),
RawTraceEventPayload::InferenceCancelled {
inference_call_id,
upstream_request_id,
partial_response_payload,
..
} => (
inference_call_id,
ExecutionStatus::Cancelled,
None,
upstream_request_id,
partial_response_payload,
),
_ => bail!("complete_inference_call received a non-terminal inference event"),
};
if !self
.rollout
.inference_calls
@@ -156,23 +196,31 @@ impl TraceReducer {
self.reduce_inference_response(wall_time_unix_ms, &inference_call_id, payload)
})
.transpose()?;
let Some(inference) = self.rollout.inference_calls.get_mut(&inference_call_id) else {
bail!("inference call {inference_call_id} disappeared during response reduction");
};
// Turn-end cleanup can close a stream before the async mapper observes
// cancellation. Preserve that terminal status while still retaining any
// late partial response evidence from the mapper.
if inference.execution.status == ExecutionStatus::Running {
inference.execution.ended_at_unix_ms = Some(wall_time_unix_ms);
inference.execution.ended_seq = Some(seq);
inference.execution.status = status;
inference.upstream_request_id = response_id;
}
if let Some(response_payload) = response_payload {
inference.raw_response_payload_id = Some(response_payload.raw_payload_id);
}
if let Some(response_item_ids) = response_item_ids {
inference.response_item_ids = response_item_ids;
{
let Some(inference) = self.rollout.inference_calls.get_mut(&inference_call_id) else {
bail!("inference call {inference_call_id} disappeared during response reduction");
};
inference.response_id = response_id;
// Turn-end cleanup can close a stream before the async mapper observes
// cancellation. Preserve that terminal status while still retaining any
// late partial response evidence from the mapper.
if inference.execution.status == ExecutionStatus::Running {
inference.execution.ended_at_unix_ms = Some(wall_time_unix_ms);
inference.execution.ended_seq = Some(seq);
inference.execution.status = status;
}
// Turn-end cleanup can mark an inference terminal before the stream
// mapper records its late partial payload. Keep the server request
// id from that late payload even when the status is already closed.
if let Some(upstream_request_id) = upstream_request_id {
inference.upstream_request_id = Some(upstream_request_id);
}
if let Some(response_payload) = response_payload {
inference.raw_response_payload_id = Some(response_payload.raw_payload_id);
}
if let Some(response_item_ids) = response_item_ids {
inference.response_item_ids = response_item_ids;
}
}
Ok(())
}
codex-rs/rollout-trace/src/reducer/inference_tests.rs
+10
View File
@@ -40,6 +40,7 @@ fn cancelled_inference_reduces_partial_response_items() -> anyhow::Result<()> {
)?;
writer.append(RawTraceEventPayload::InferenceCancelled {
inference_call_id: "inference-1".to_string(),
upstream_request_id: Some("req-cancelled".to_string()),
reason: "test interruption".to_string(),
partial_response_payload: Some(partial_response),
})?;
@@ -49,6 +50,10 @@ fn cancelled_inference_reduces_partial_response_items() -> anyhow::Result<()> {
let response_item_id = &inference.response_item_ids[0];
assert_eq!(inference.execution.status, ExecutionStatus::Cancelled);
assert_eq!(
inference.upstream_request_id,
Some("req-cancelled".to_string()),
);
assert_eq!(inference.response_item_ids.len(), 1);
assert_eq!(
rollout.conversation_items[response_item_id].kind,
@@ -123,6 +128,7 @@ fn late_cancelled_inference_preserves_turn_end_status() -> anyhow::Result<()> {
)?;
writer.append(RawTraceEventPayload::InferenceCancelled {
inference_call_id: "inference-1".to_string(),
upstream_request_id: Some("req-late-cancelled".to_string()),
reason: "stream mapper noticed cancellation after turn end".to_string(),
partial_response_payload: Some(partial_response.clone()),
})?;
@@ -135,6 +141,10 @@ fn late_cancelled_inference_preserves_turn_end_status() -> anyhow::Result<()> {
inference.raw_response_payload_id,
Some(partial_response.raw_payload_id),
);
assert_eq!(
inference.upstream_request_id,
Some("req-late-cancelled".to_string()),
);
assert_eq!(inference.response_item_ids.len(), 1);
let response_item_id = &inference.response_item_ids[0];
assert_eq!(
codex-rs/rollout-trace/src/reducer/mod.rs
+4 -41
View File
@@ -226,47 +226,10 @@ impl TraceReducer {
},
)?;
}
RawTraceEventPayload::InferenceCompleted {
inference_call_id,
response_id,
response_payload,
} => {
self.complete_inference_call(
event.seq,
event.wall_time_unix_ms,
inference_call_id,
ExecutionStatus::Completed,
response_id,
Some(response_payload),
)?;
}
RawTraceEventPayload::InferenceFailed {
inference_call_id,
partial_response_payload,
..
} => {
self.complete_inference_call(
event.seq,
event.wall_time_unix_ms,
inference_call_id,
ExecutionStatus::Failed,
/*response_id*/ None,
partial_response_payload,
)?;
}
RawTraceEventPayload::InferenceCancelled {
inference_call_id,
partial_response_payload,
..
} => {
self.complete_inference_call(
event.seq,
event.wall_time_unix_ms,
inference_call_id,
ExecutionStatus::Cancelled,
/*response_id*/ None,
partial_response_payload,
)?;
payload @ (RawTraceEventPayload::InferenceCompleted { .. }
| RawTraceEventPayload::InferenceFailed { .. }
| RawTraceEventPayload::InferenceCancelled { .. }) => {
self.complete_inference_call(event.seq, event.wall_time_unix_ms, payload)?;
}
RawTraceEventPayload::ProtocolEventObserved { .. } => {
// Protocol wrappers are raw debug breadcrumbs. Typed hooks own
codex-rs/rollout-trace/src/reducer/test_support.rs
+2
View File
@@ -146,6 +146,7 @@ pub(crate) fn append_inference_completion(
writer.append(RawTraceEventPayload::InferenceCompleted {
inference_call_id: inference_call_id.to_string(),
response_id: Some(response_id.to_string()),
upstream_request_id: None,
response_payload,
})?;
Ok(())
@@ -184,6 +185,7 @@ pub(crate) fn append_completed_inference(
RawTraceEventPayload::InferenceCompleted {
inference_call_id: inference_id.to_string(),
response_id: Some(format!("resp-{inference_id}")),
upstream_request_id: None,
response_payload: response,
},
)?;
codex-rs/rollout-trace/src/reducer/tool/terminal_tests.rs
+1
View File
@@ -551,6 +551,7 @@ fn append_inference_with_tool_call(writer: &TraceWriter) -> anyhow::Result<()> {
writer.append(RawTraceEventPayload::InferenceCompleted {
inference_call_id: "inference-1".to_string(),
response_id: Some("resp-1".to_string()),
upstream_request_id: None,
response_payload: response,
})?;
Ok(())
codex-rs/rollout-trace/src/thread_tests.rs
+2 -2
View File
@@ -133,8 +133,8 @@ fn disabled_thread_context_accepts_trace_calls_without_writing() -> anyhow::Resu
let inference_attempt = inference_trace.start_attempt();
inference_attempt.record_started(&serde_json::json!({ "kind": "inference" }));
let token_usage: Option<codex_protocol::protocol::TokenUsage> = None;
inference_attempt.record_completed("response-1", &token_usage, &[]);
inference_attempt.record_failed("inference failed", &[]);
inference_attempt.record_completed("response-1", Some("req-1"), &token_usage, &[]);
inference_attempt.record_failed("inference failed", /*upstream_request_id*/ None, &[]);
let compaction_trace = thread_trace.compaction_trace_context(
"turn-1",
codex-rs/rollout-trace/src/writer.rs
+1
View File
@@ -226,6 +226,7 @@ mod tests {
writer.append(RawTraceEventPayload::InferenceCompleted {
inference_call_id: "inference-1".to_string(),
response_id: Some("resp-1".to_string()),
upstream_request_id: Some("req-1".to_string()),
response_payload: inference_response.clone(),
})?;
writer.append(RawTraceEventPayload::CodexTurnEnded {