i/Aether
1
0
Fork 0
mirror of https://github.com/fawney19/Aether.git synced 2026-10-04 16:37:46 +08:00
Code Issues Packages Projects Releases Wiki Activity

feat(health): refine health monitor metrics

Browse Source 
- add TPS to model and provider health payloads

- exclude user-cancelled 499 requests from health statistics

- update model/provider health cards with average latency, average TTFB, TPS, and availability
This commit is contained in:
AAEE86
2026-06-01 18:34:51 +08:00
parent 6c2a57f237
commit 7ffe33f867
11 changed files with 366 additions and 172 deletions
Download Patch File Download Diff File Expand all files Collapse all files
apps/aether-gateway/src/handlers/public/catalog_helpers.rs
+20 -2
View File
@@ -23,6 +23,8 @@ use serde_json::json;
use std::collections::{BTreeMap, BTreeSet};
use std::time::{SystemTime, UNIX_EPOCH};
const USER_CANCELLED_STATUS_CODE: u16 = 499;
pub(crate) fn request_candidate_status_label(status: RequestCandidateStatus) -> &'static str {
match status {
RequestCandidateStatus::Available => "available",
@@ -576,6 +578,7 @@ pub(crate) async fn build_model_health_monitor_payload(
created_until_unix_secs: now_unix_secs,
user_id: None,
provider_name: None,
exclude_status_codes: vec![USER_CANCELLED_STATUS_CODE],
group_by: UsageBreakdownGroupBy::Model,
})
.await
@@ -595,6 +598,7 @@ pub(crate) async fn build_model_health_monitor_payload(
created_from_unix_secs: Some(since_unix_secs),
created_until_unix_secs: Some(now_unix_secs),
model: Some(row.group_key.clone()),
exclude_status_codes: vec![USER_CANCELLED_STATUS_CODE],
limit: Some(per_model_limit),
newest_first: true,
..UsageAuditListQuery::default()
@@ -640,6 +644,7 @@ pub(crate) async fn build_model_health_monitor_payload(
"success_rate": success_rate,
"avg_latency_ms": avg_latency_ms,
"avg_first_byte_ms": first_byte_average,
"avg_tps": model_health_average_tps(&row),
"last_event_at": last_event_at,
"events": event_payload,
"timeline": timeline,
@@ -691,6 +696,7 @@ pub(crate) async fn build_provider_health_monitor_payload(
created_until_unix_secs: now_unix_secs,
user_id: None,
provider_name: None,
exclude_status_codes: vec![USER_CANCELLED_STATUS_CODE],
group_by: UsageBreakdownGroupBy::Provider,
})
.await
@@ -754,6 +760,7 @@ async fn build_provider_health_payload(
created_until_unix_secs: now_unix_secs,
user_id: None,
provider_name: Some(provider.name.clone()),
exclude_status_codes: vec![USER_CANCELLED_STATUS_CODE],
group_by: UsageBreakdownGroupBy::Model,
})
.await
@@ -768,6 +775,7 @@ async fn build_provider_health_payload(
created_from_unix_secs: Some(since_unix_secs),
created_until_unix_secs: Some(now_unix_secs),
provider_name: Some(provider.name.clone()),
exclude_status_codes: vec![USER_CANCELLED_STATUS_CODE],
limit: Some(provider_event_limit),
newest_first: true,
..UsageAuditListQuery::default()
@@ -804,7 +812,7 @@ async fn build_provider_health_payload(
));
}
let (total_attempts, success_count, failed_count, success_rate, avg_latency_ms) =
let (total_attempts, success_count, failed_count, success_rate, avg_latency_ms, avg_tps) =
if let Some(row) = provider_stats {
let total_attempts = row.request_count;
let success_count = row.success_count.min(total_attempts);
@@ -820,9 +828,10 @@ async fn build_provider_health_payload(
failed_count,
success_rate,
model_health_average_latency_ms(row),
model_health_average_tps(row),
)
} else {
(0, 0, 0, 1.0, None)
(0, 0, 0, 1.0, None, None)
};
let (timeline, time_range_start, time_range_end) = build_model_health_timeline(
@@ -847,6 +856,7 @@ async fn build_provider_health_payload(
"success_rate": success_rate,
"avg_latency_ms": avg_latency_ms,
"avg_first_byte_ms": model_health_average_first_byte_ms(&provider_events),
"avg_tps": avg_tps,
"model_count": model_breakdown.len(),
"last_event_at": last_event_at,
"timeline": timeline,
@@ -917,6 +927,7 @@ fn model_health_payload_from_row(
"success_rate": success_rate,
"avg_latency_ms": model_health_average_latency_ms(row),
"avg_first_byte_ms": model_health_average_first_byte_ms(events),
"avg_tps": model_health_average_tps(row),
"last_event_at": last_event_at,
"events": event_payload,
"timeline": timeline,
@@ -939,6 +950,13 @@ fn model_health_average_latency_ms(row: &StoredUsageBreakdownSummaryRow) -> Opti
None
}
fn model_health_average_tps(row: &StoredUsageBreakdownSummaryRow) -> Option<f64> {
if row.output_tokens == 0 || row.response_time_sum_ms <= 0.0 {
return None;
}
Some(row.output_tokens as f64 / (row.response_time_sum_ms / 1000.0))
}
fn model_health_average_first_byte_ms(events: &[StoredRequestUsageAudit]) -> Option<f64> {
let mut sum = 0u64;
let mut count = 0u64;
apps/aether-gateway/src/handlers/public/support/user_me_usage.rs
+7
View File
@@ -848,6 +848,7 @@ pub(super) async fn handle_users_me_usage_get(
created_until_unix_secs,
user_id: Some(auth.user.id.clone()),
provider_name: None,
exclude_status_codes: Vec::new(),
group_by: UsageBreakdownGroupBy::Model,
})
.await
@@ -868,6 +869,7 @@ pub(super) async fn handle_users_me_usage_get(
created_until_unix_secs,
user_id: Some(auth.user.id.clone()),
provider_name: None,
exclude_status_codes: Vec::new(),
group_by: UsageBreakdownGroupBy::Provider,
})
.await
@@ -888,6 +890,7 @@ pub(super) async fn handle_users_me_usage_get(
created_until_unix_secs,
user_id: Some(auth.user.id.clone()),
provider_name: None,
exclude_status_codes: Vec::new(),
group_by: UsageBreakdownGroupBy::ApiFormat,
})
.await
@@ -926,6 +929,7 @@ pub(super) async fn handle_users_me_usage_get(
model: None,
api_format: None,
statuses: None,
exclude_status_codes: Vec::new(),
is_stream: None,
error_only: false,
keywords,
@@ -979,6 +983,7 @@ pub(super) async fn handle_users_me_usage_get(
model: None,
api_format: None,
statuses: None,
exclude_status_codes: Vec::new(),
is_stream: None,
error_only: false,
limit: None,
@@ -1005,6 +1010,7 @@ pub(super) async fn handle_users_me_usage_get(
model: None,
api_format: None,
statuses: None,
exclude_status_codes: Vec::new(),
is_stream: None,
error_only: false,
limit: Some(limit),
@@ -1141,6 +1147,7 @@ pub(super) async fn handle_users_me_usage_active_get(
model: None,
api_format: None,
statuses: Some(vec!["pending".to_string(), "streaming".to_string()]),
exclude_status_codes: Vec::new(),
is_stream: None,
error_only: false,
limit: Some(50),