fix(api): save call logs and add endpoint fallback for local rerank providers (#11081)

Validated on a combined board over tip aa128736: focused tests local-rerank-logging + call-logs-row-filter green (7/7 across runs), typecheck:core clean. One pre-merge fix: widened waitForCallLogSaves 5s→15s — the call-log artifact writer's fsync latency exceeds 5s under a loaded host (pre-existing call-log-save-drain flakes identically on the pure tip), so the budget was load-flaky, not the code. Local rerank now logs call entries on success and error paths like embeddings/cloud rerank, captures full request/response payloads, attaches x-omniroute-* headers, and falls back /v1/rerank→/rerank on 404. Thank you @AndrianBalanescu!
This commit is contained in:
Andrew B.
2026-08-22 14:54:17 -05:00
committed by GitHub
parent aa12873686
commit 56540f24c5
3 changed files with 345 additions and 14 deletions

View File

@@ -10,11 +10,15 @@ import { HTTP_STATUS } from "@omniroute/open-sse/config/constants.ts";
import { enforceApiKeyPolicy } from "@/shared/utils/apiKeyPolicy";
import { v1RerankSchema } from "@/shared/validation/schemas";
import { isValidationFailure, validateBody } from "@/shared/validation/helpers";
import { getCachedProviderNodes } from "@/lib/localDb";
import { getCachedProviderNodes } from "@/lib/db/readCache";
import {
isAllRateLimitedCredentials,
rateLimitedProviderResponse,
} from "@/app/api/v1/_shared/rateLimit";
import { saveCallLog } from "@/lib/usageDb";
import { attachOmniRouteMetaHeaders } from "@/domain/omnirouteResponseMeta";
import { generateRequestId } from "@/shared/utils/requestId";
import { CORS_HEADERS } from "@omniroute/open-sse/utils/cors.ts";
/**
* Handle CORS preflight
@@ -121,6 +125,8 @@ async function postHandler(request, context) {
return_documents: body.return_documents,
credentials,
connectionId: (credentials as { connectionId?: string } | null)?.connectionId || null,
apiKeyId: policy.apiKeyInfo?.id || null,
apiKeyName: policy.apiKeyInfo?.name || null,
});
if (response?.ok) {
await clearRecoveredProviderState(credentials);
@@ -148,8 +154,9 @@ async function postHandler(request, context) {
}
const token = credentials?.apiKey || credentials?.accessToken;
const startTime = Date.now();
try {
const res = await fetch(localProvider.baseUrl, {
let res = await fetch(localProvider.baseUrl, {
method: "POST",
headers: {
"Content-Type": "application/json",
@@ -164,19 +171,110 @@ async function postHandler(request, context) {
}),
});
// Some local providers (e.g. Infinity, TEI) mount at /rerank rather than /v1/rerank
if (res.status === 404 && localProvider.baseUrl.endsWith("/v1/rerank")) {
const fallbackUrl = localProvider.baseUrl.replace(/\/v1\/rerank$/, "/rerank");
try {
const fallbackRes = await fetch(fallbackUrl, {
method: "POST",
headers: {
"Content-Type": "application/json",
Authorization: `Bearer ${token}`,
},
body: JSON.stringify({
model: localModel,
query: body.query,
documents: body.documents,
top_n: body.top_n || body.documents.length,
return_documents: body.return_documents !== false,
}),
});
if (fallbackRes.ok || fallbackRes.status !== 404) {
res = fallbackRes;
}
} catch {
// retain original 404 response if fallback fetch fails
}
}
if (!res.ok) {
const errData = await res.json().catch(() => ({}));
return errorResponse(
res.status,
errData.message || errData.detail || `Provider returned HTTP ${res.status}`
);
const errorMessage =
errData.message || errData.detail || `Provider returned HTTP ${res.status}`;
saveCallLog({
method: "POST",
path: "/v1/rerank",
status: res.status,
model: body.model,
provider: prefix,
connectionId:
(credentials as { connectionId?: string } | null)?.connectionId || undefined,
duration: Date.now() - startTime,
requestBody: {
model: body.model,
query: body.query,
documents: body.documents,
top_n: body.top_n,
return_documents: body.return_documents,
},
responseBody: errData,
error: errorMessage,
apiKeyId: policy.apiKeyInfo?.id || undefined,
apiKeyName: policy.apiKeyInfo?.name || undefined,
}).catch(() => {});
return errorResponse(res.status, errorMessage);
}
const data = await res.json();
return Response.json(data, {
headers: {},
const latencyMs = Date.now() - startTime;
saveCallLog({
method: "POST",
path: "/v1/rerank",
status: 200,
model: body.model,
provider: prefix,
connectionId:
(credentials as { connectionId?: string } | null)?.connectionId || undefined,
duration: latencyMs,
tokens: { prompt_tokens: 0, completion_tokens: 0 },
requestBody: {
model: body.model,
query: body.query,
documents: body.documents,
top_n: body.top_n,
return_documents: body.return_documents,
},
responseBody: data,
apiKeyId: policy.apiKeyInfo?.id || undefined,
apiKeyName: policy.apiKeyInfo?.name || undefined,
}).catch(() => {});
const headers = new Headers({ ...CORS_HEADERS, "Content-Type": "application/json" });
attachOmniRouteMetaHeaders(headers, {
provider: prefix,
model: localModel,
costUsd: 0,
latencyMs,
requestId: generateRequestId(),
});
return new Response(JSON.stringify(data), {
status: 200,
headers,
});
} catch (err: any) {
saveCallLog({
method: "POST",
path: "/v1/rerank",
status: 500,
model: body.model,
provider: prefix,
connectionId:
(credentials as { connectionId?: string } | null)?.connectionId || undefined,
duration: Date.now() - startTime,
error: err.message,
apiKeyId: policy.apiKeyInfo?.id || undefined,
apiKeyName: policy.apiKeyInfo?.name || undefined,
}).catch(() => {});
return errorResponse(500, `Rerank request failed: ${err.message}`);
}
}