[codex] Add turn profiling analytics (#26484)

## Summary

Add flat profiling fields to `codex_turn_event` so analytics can explain
where turn wall-clock time is spent without changing tool execution
behavior.

The profile reports:
- time before the first sampling request
- sampling time across all attempts and follow-ups
- overhead between sampling requests
- time blocked in the post-sampling tool drain
- time after the final sampling request
- sampling request and retry counts

## Implementation

- Extend the existing turn timing state with constant-memory phase
accounting and one RAII phase guard.
- Observe sampling and the existing post-sampling drain only at turn
orchestration boundaries.
- Keep tool runtime, tool futures, response item handling, and turn
lifecycle values unchanged.
- Add the profiling fields directly to the existing analytics turn event
without changing app-server protocol or rollout persistence.
- Use the existing turn `status` to distinguish completed, failed, and
interrupted profiles.

Exact sampling/tool overlap is intentionally omitted because measuring
tool completion accurately would require hooks in the tool execution
path.

## Validation

- Add app-server end-to-end coverage for a single-sampling turn with no
blocking tool work.
- Add app-server end-to-end coverage for `request_user_input` blocking
followed by a second sampling request.
- CI is running on the PR; tests were not executed locally per
repository guidance.
This commit is contained in:
Ahmed Ibrahim
2026-06-05 11:27:10 -07:00
committed by GitHub
Unverified
parent 82b15b65e2
commit 8d72fb6de9
11 changed files with 530 additions and 101 deletions
@@ -63,6 +63,8 @@ use crate::facts::SubAgentThreadStartedInput;
use crate::facts::ThreadInitializationMode;
use crate::facts::TrackEventsContext;
use crate::facts::TurnCodexErrorFact;
use crate::facts::TurnProfile;
use crate::facts::TurnProfileFact;
use crate::facts::TurnResolvedConfigFact;
use crate::facts::TurnStatus;
use crate::facts::TurnSteerRequestError;
@@ -396,6 +398,18 @@ fn sample_turn_resolved_config(thread_id: &str, turn_id: &str) -> TurnResolvedCo
}
}
fn sample_turn_profile() -> TurnProfile {
TurnProfile {
before_first_sampling_ms: 100,
sampling_ms: 700,
between_sampling_overhead_ms: 50,
tool_blocking_ms: 250,
after_last_sampling_ms: 134,
sampling_request_count: 2,
sampling_retry_count: 1,
}
}
fn sample_turn_steer_request(
thread_id: &str,
expected_turn_id: &str,
@@ -649,6 +663,18 @@ async fn ingest_turn_prerequisites(
)
.await;
}
reducer
.ingest(
AnalyticsFact::Custom(CustomAnalyticsFact::TurnProfile(Box::new(
TurnProfileFact {
turn_id: "turn-2".to_string(),
profile: sample_turn_profile(),
},
))),
out,
)
.await;
}
async fn ingest_review_prerequisites(
@@ -3300,6 +3326,13 @@ fn turn_event_serializes_expected_shape() {
output_tokens: None,
reasoning_output_tokens: None,
total_tokens: None,
before_first_sampling_ms: 100,
sampling_ms: 700,
between_sampling_overhead_ms: 50,
tool_blocking_ms: 250,
after_last_sampling_ms: 134,
sampling_request_count: 2,
sampling_retry_count: 1,
duration_ms: Some(1234),
started_at: Some(455),
completed_at: Some(456),
@@ -3366,6 +3399,13 @@ fn turn_event_serializes_expected_shape() {
"output_tokens": null,
"reasoning_output_tokens": null,
"total_tokens": null,
"before_first_sampling_ms": 100,
"sampling_ms": 700,
"between_sampling_overhead_ms": 50,
"tool_blocking_ms": 250,
"after_last_sampling_ms": 134,
"sampling_request_count": 2,
"sampling_retry_count": 1,
"duration_ms": 1234,
"started_at": 455,
"completed_at": 456
+7
View File
@@ -19,6 +19,7 @@ use crate::facts::SkillInvokedInput;
use crate::facts::SubAgentThreadStartedInput;
use crate::facts::TrackEventsContext;
use crate::facts::TurnCodexErrorFact;
use crate::facts::TurnProfileFact;
use crate::facts::TurnResolvedConfigFact;
use crate::facts::TurnTokenUsageFact;
use crate::reducer::AnalyticsReducer;
@@ -257,6 +258,12 @@ impl AnalyticsEventsClient {
)));
}
pub fn track_turn_profile(&self, fact: TurnProfileFact) {
self.record_fact(AnalyticsFact::Custom(CustomAnalyticsFact::TurnProfile(
Box::new(fact),
)));
}
pub fn track_turn_codex_error(&self, fact: TurnCodexErrorFact) {
self.record_fact(AnalyticsFact::Custom(CustomAnalyticsFact::TurnCodexError(
Box::new(fact),
+7
View File
@@ -817,6 +817,13 @@ pub(crate) struct CodexTurnEventParams {
pub(crate) output_tokens: Option<i64>,
pub(crate) reasoning_output_tokens: Option<i64>,
pub(crate) total_tokens: Option<i64>,
pub(crate) before_first_sampling_ms: u64,
pub(crate) sampling_ms: u64,
pub(crate) between_sampling_overhead_ms: u64,
pub(crate) tool_blocking_ms: u64,
pub(crate) after_last_sampling_ms: u64,
pub(crate) sampling_request_count: u32,
pub(crate) sampling_retry_count: u32,
pub(crate) duration_ms: Option<u64>,
pub(crate) started_at: Option<u64>,
pub(crate) completed_at: Option<u64>,
+18
View File
@@ -104,6 +104,23 @@ pub struct TurnTokenUsageFact {
pub token_usage: TokenUsage,
}
#[derive(Clone, Debug, PartialEq, Eq, Serialize)]
pub struct TurnProfile {
pub before_first_sampling_ms: u64,
pub sampling_ms: u64,
pub between_sampling_overhead_ms: u64,
pub tool_blocking_ms: u64,
pub after_last_sampling_ms: u64,
pub sampling_request_count: u32,
pub sampling_retry_count: u32,
}
#[derive(Clone)]
pub struct TurnProfileFact {
pub turn_id: String,
pub profile: TurnProfile,
}
#[derive(Clone)]
pub struct TurnCodexErrorFact {
pub(crate) turn_id: String,
@@ -476,6 +493,7 @@ pub(crate) enum CustomAnalyticsFact {
GuardianReview(Box<GuardianReviewEventParams>),
TurnResolvedConfig(Box<TurnResolvedConfigFact>),
TurnTokenUsage(Box<TurnTokenUsageFact>),
TurnProfile(Box<TurnProfileFact>),
TurnCodexError(Box<TurnCodexErrorFact>),
SkillInvoked(SkillInvokedInput),
AppMentioned(AppMentionedInput),
+2
View File
@@ -39,6 +39,8 @@ pub use facts::SubAgentThreadStartedInput;
pub use facts::ThreadInitializationMode;
pub use facts::TrackEventsContext;
pub use facts::TurnCodexErrorFact;
pub use facts::TurnProfile;
pub use facts::TurnProfileFact;
pub use facts::TurnResolvedConfigFact;
pub use facts::TurnStatus;
pub use facts::TurnSteerRejectionReason;
+52 -99
View File
@@ -72,6 +72,8 @@ use crate::facts::SubAgentThreadStartedInput;
use crate::facts::ThreadInitializationMode;
use crate::facts::TurnCodexError;
use crate::facts::TurnCodexErrorFact;
use crate::facts::TurnProfile;
use crate::facts::TurnProfileFact;
use crate::facts::TurnResolvedConfigFact;
use crate::facts::TurnStatus;
use crate::facts::TurnSteerRejectionReason;
@@ -316,6 +318,7 @@ struct CompletedTurnState {
duration_ms: Option<u64>,
}
#[derive(Default)]
struct TurnState {
connection_id: Option<u64>,
thread_id: Option<String>,
@@ -323,6 +326,7 @@ struct TurnState {
resolved_config: Option<TurnResolvedConfigFact>,
started_at: Option<u64>,
token_usage: Option<TokenUsage>,
profile: Option<TurnProfile>,
completed: Option<CompletedTurnState>,
codex_error: Option<TurnCodexError>,
latest_diff: Option<String>,
@@ -464,6 +468,9 @@ impl AnalyticsReducer {
CustomAnalyticsFact::TurnTokenUsage(input) => {
self.ingest_turn_token_usage(*input, out).await;
}
CustomAnalyticsFact::TurnProfile(input) => {
self.ingest_turn_profile(*input, out).await;
}
CustomAnalyticsFact::TurnCodexError(input) => {
self.ingest_turn_codex_error(*input);
}
@@ -604,19 +611,7 @@ impl AnalyticsReducer {
let turn_id = input.turn_id.clone();
let thread_id = input.thread_id.clone();
let num_input_images = input.num_input_images;
let turn_state = self.turns.entry(turn_id.clone()).or_insert(TurnState {
connection_id: None,
thread_id: None,
num_input_images: None,
resolved_config: None,
started_at: None,
token_usage: None,
completed: None,
codex_error: None,
latest_diff: None,
steer_count: 0,
tool_counts: TurnToolCounts::default(),
});
let turn_state = self.turns.entry(turn_id.clone()).or_default();
turn_state.thread_id = Some(thread_id);
turn_state.num_input_images = Some(num_input_images);
turn_state.resolved_config = Some(input);
@@ -629,43 +624,30 @@ impl AnalyticsReducer {
out: &mut Vec<TrackEventRequest>,
) {
let turn_id = input.turn_id.clone();
let turn_state = self.turns.entry(turn_id.clone()).or_insert(TurnState {
connection_id: None,
thread_id: None,
num_input_images: None,
resolved_config: None,
started_at: None,
token_usage: None,
completed: None,
codex_error: None,
latest_diff: None,
steer_count: 0,
tool_counts: TurnToolCounts::default(),
});
let turn_state = self.turns.entry(turn_id.clone()).or_default();
turn_state.thread_id = Some(input.thread_id);
turn_state.token_usage = Some(input.token_usage);
self.maybe_emit_turn_event(&turn_id, out).await;
}
async fn ingest_turn_profile(
&mut self,
input: TurnProfileFact,
out: &mut Vec<TrackEventRequest>,
) {
let TurnProfileFact { turn_id, profile } = input;
let turn_state = self.turns.entry(turn_id.clone()).or_default();
turn_state.profile = Some(profile);
self.maybe_emit_turn_event(&turn_id, out).await;
}
fn ingest_turn_codex_error(&mut self, input: TurnCodexErrorFact) {
let TurnCodexErrorFact {
turn_id,
thread_id,
error,
} = input;
let turn_state = self.turns.entry(turn_id).or_insert(TurnState {
connection_id: None,
thread_id: None,
num_input_images: None,
resolved_config: None,
started_at: None,
token_usage: None,
completed: None,
codex_error: None,
latest_diff: None,
steer_count: 0,
tool_counts: TurnToolCounts::default(),
});
let turn_state = self.turns.entry(turn_id).or_default();
turn_state.thread_id.get_or_insert(thread_id);
turn_state.codex_error = Some(error);
}
@@ -818,19 +800,7 @@ impl AnalyticsReducer {
else {
return;
};
let turn_state = self.turns.entry(turn_id.clone()).or_insert(TurnState {
connection_id: None,
thread_id: None,
num_input_images: None,
resolved_config: None,
started_at: None,
token_usage: None,
completed: None,
codex_error: None,
latest_diff: None,
steer_count: 0,
tool_counts: TurnToolCounts::default(),
});
let turn_state = self.turns.entry(turn_id.clone()).or_default();
turn_state.connection_id = Some(connection_id);
turn_state.thread_id = Some(pending_request.thread_id);
turn_state.num_input_images = Some(pending_request.num_input_images);
@@ -1178,61 +1148,19 @@ impl AnalyticsReducer {
self.ingest_guardian_review_completed(notification, out);
}
ServerNotification::TurnStarted(notification) => {
let turn_state = self.turns.entry(notification.turn.id).or_insert(TurnState {
connection_id: None,
thread_id: None,
num_input_images: None,
resolved_config: None,
started_at: None,
token_usage: None,
completed: None,
codex_error: None,
latest_diff: None,
steer_count: 0,
tool_counts: TurnToolCounts::default(),
});
let turn_state = self.turns.entry(notification.turn.id).or_default();
turn_state.started_at = notification
.turn
.started_at
.and_then(|started_at| u64::try_from(started_at).ok());
}
ServerNotification::TurnDiffUpdated(notification) => {
let turn_state =
self.turns
.entry(notification.turn_id.clone())
.or_insert(TurnState {
connection_id: None,
thread_id: None,
num_input_images: None,
resolved_config: None,
started_at: None,
token_usage: None,
completed: None,
codex_error: None,
latest_diff: None,
steer_count: 0,
tool_counts: TurnToolCounts::default(),
});
let turn_state = self.turns.entry(notification.turn_id.clone()).or_default();
turn_state.thread_id = Some(notification.thread_id);
turn_state.latest_diff = Some(notification.diff);
}
ServerNotification::TurnCompleted(notification) => {
let turn_state =
self.turns
.entry(notification.turn.id.clone())
.or_insert(TurnState {
connection_id: None,
thread_id: None,
num_input_images: None,
resolved_config: None,
started_at: None,
token_usage: None,
completed: None,
codex_error: None,
latest_diff: None,
steer_count: 0,
tool_counts: TurnToolCounts::default(),
});
let turn_state = self.turns.entry(notification.turn.id.clone()).or_default();
turn_state.completed = Some(CompletedTurnState {
status: analytics_turn_status(notification.turn.status),
turn_error: notification
@@ -1511,6 +1439,7 @@ impl AnalyticsReducer {
if turn_state.thread_id.is_none()
|| turn_state.num_input_images.is_none()
|| turn_state.resolved_config.is_none()
|| turn_state.profile.is_none()
|| turn_state.completed.is_none()
{
return;
@@ -2457,12 +2386,20 @@ fn codex_turn_event_params(
turn_state: &TurnState,
thread_metadata: &ThreadMetadataState,
) -> CodexTurnEventParams {
let (Some(thread_id), Some(num_input_images), Some(resolved_config), Some(completed)) = (
let (
Some(thread_id),
Some(num_input_images),
Some(resolved_config),
Some(profile),
Some(completed),
) = (
turn_state.thread_id.clone(),
turn_state.num_input_images,
turn_state.resolved_config.clone(),
turn_state.profile.clone(),
turn_state.completed.clone(),
) else {
)
else {
unreachable!("turn event params require a fully populated turn state");
};
let started_at = turn_state.started_at;
@@ -2488,6 +2425,15 @@ fn codex_turn_event_params(
workspace_kind,
is_first_turn,
} = resolved_config;
let TurnProfile {
before_first_sampling_ms,
sampling_ms,
between_sampling_overhead_ms,
tool_blocking_ms,
after_last_sampling_ms,
sampling_request_count,
sampling_retry_count,
} = profile;
let token_usage = turn_state.token_usage.clone();
let codex_error = turn_state.codex_error.as_ref();
CodexTurnEventParams {
@@ -2550,6 +2496,13 @@ fn codex_turn_event_params(
total_tokens: token_usage
.as_ref()
.map(|token_usage| token_usage.total_tokens),
before_first_sampling_ms,
sampling_ms,
between_sampling_overhead_ms,
tool_blocking_ms,
after_last_sampling_ms,
sampling_request_count,
sampling_retry_count,
duration_ms: completed.duration_ms,
started_at,
completed_at: Some(completed.completed_at),