mslxdff 0.1.36 → 0.1.38

package/bin/mslxdff.js

@@ -1179,17 +1179,44 @@ function fmtEvent(e) {
   const t = e?.ts ? new Date(e.ts).toISOString().slice(11, 19) : "--:--:--";
   const head = `[${t}]`;
   const m = (x) => x || "-";
+  const fallbackTag = e.fallback ? ` fallback=${e.fallback.requested_model || e.requested}->${e.fallback.actual_model || e.actual}(${e.fallback.reason || e.reason})` : "";
   switch (e?.type) {
     case "request":
-      return `${head} request       ${m(e.model)}${e.auto ? " (auto)" : ""} hops=${e.hops} from ${e.ip || "?"}${e.stream ? " stream" : ""}${e.prompt ? ` content="${e.prompt}"` : ""}`;
+      return `${head} request       client-> ${m(e.requested || e.model)}${e.auto ? " (auto)" : ""} raw=${m(e.rawModel)} lock=${m(e.lockModel)} hops=${e.hops} from ${e.ip || "?"}${e.stream ? " stream" : ""}${e.prompt ? ` content="${e.prompt}"` : ""} reqId=${e.reqId || ""}`;
+    case "client-request":
+      return `${head} client req    客户端请求 model=${m(e.requested)} raw=${m(e.rawModel)} lock=${m(e.lockModel)} ip=${e.ip || "?"}${e.stream ? " stream" : ""}`;
+    case "ordered":
+      return `${head} ordered       尝试顺序 [${(e.order||[]).join(" -> ")}] canFallback=${e.canFallback} useAuto=${e.useAuto}`;
+    case "model-try":
+      return `${head} model-try     尝试本地 model=${m(e.model)} idx=${e.idx} reqId=${e.reqId || ""}`;
+    case "upstream-try":
+      return `${head} upstream try  上游请求 model=${m(e.model)} attempt=${e.attempt || 1}`;
+    case "upstream-done":
+      return `${head} upstream ok   ${m(e.model)} HTTP ${e.status}${e.timing ? ` total=${e.timing.totalMs}ms` : ""}`;
     case "upstream-error":
       return `${head} upstream err  ${m(e.model)} ${e.status ? `HTTP ${e.status}` : "network"}: ${m(e.message)}`;
+    case "peer-race-start":
+      return `${head} peer race     开始并发给组员 model=${m(e.model)} peers=${e.peers}`;
     case "peer-health":
-      return `${head} peer check    ${e.peer} -> ${e.count ? e.healthy.join(", ") : "no healthy models"}`;
+      return `${head} peer check    ${e.peer} -> ${e.count ? e.healthy.join(", ") : "no healthy models"}${e.strict ? " (strict)" : ""}`;
+    case "peer-request":
+      return `${head} peer req      并发发出 peer=${e.peer} model=${m(e.model)} hops=${e.hops}`;
     case "peer-forward":
-      return `${head} forward ->    ${e.peer} model=${m(e.model)} hops=${e.hops}${e.retry ? " (retry)" : ""}`;
+      return `${head} peer resp     收到响应 peer=${e.peer} model=${m(e.model)} ${e.ok ? "ok" : "fail"} ${e.latencyMs ? `${e.latencyMs}ms` : ""} hops=${e.hops}${e.retry ? " (retry)" : ""}`;
     case "peer-error":
-      return `${head} peer err      ${e.peer} ${e.status ? `HTTP ${e.status}` : "network"}: ${m(e.message)}`;
+      return `${head} peer err      ${e.peer} ${e.status ? `HTTP ${e.status}` : "network"}: ${m(e.message)} model=${m(e.model)}`;
+    case "peer-race-win":
+      return `${head} peer win      选中 peer=${e.winPeer} model=${m(e.winTarget)} latency=${e.latencyMs}ms`;
+    case "peer-race-lose":
+      return `${head} peer lose     组员全部失败 model=${m(e.model)}`;
+    case "fallback":
+      return `${head} fallback      ${m(e.from)} -> ${m(e.to)} reason=${m(e.reason)}`;
+    case "fallback-notice":
+      return `${head} fallback!     客户端请求 ${m(e.requested)} 实际返回 ${m(e.actual)} 原因=${m(e.reason)} via=${m(e.via)} notice=${m(e.notice)}`;
+    case "relay-try":
+      return `${head} relay try     给宽带中继请求 target=${e.target} model=${m(e.model)} via=${e.via}`;
+    case "relay-fail":
+      return `${head} relay fail    ${e.target} ${e.status ? `HTTP ${e.status}` : ""} ${m(e.message)}`;
     case "relay-ip-change":
       return `${head} relay ip      ${e.member || e.id || "?"} ${e.oldIp || "?"} -> ${e.newIp || e.publicIp || "?"} via leader`;
     case "relay-forward":
@@ -1199,7 +1226,9 @@ function fmtEvent(e) {
     case "client-abort":
       return `${head} client abort  ${m(e.model)} total=${fmtDur(e.totalMs)}`;
     case "result":
-      return `${head} result        ${e.status} ${m(e.model)} via=${e.via} 响应耗时 ${fmtDur(e.durationMs)}`;
+      return `${head} result        返回客户端 status=${e.status} model=${m(e.model)} via=${e.via} 响应耗时 ${fmtDur(e.durationMs)}${fallbackTag}`;
+    case "client-response":
+      return `${head} client res    返回客户端 客户端请求 ${m(e.requested)} 实际返回 ${m(e.actual)} via=${m(e.via)}${e.fallback ? ` fallback=${e.fallback.requested_model}->${e.fallback.actual_model}(${e.fallback.reason})` : " 无fallback"} status=${e.status}`;
     case "auto-update-enabled":
       return `${head} auto-update   enabled every ${Math.round((e.intervalMs||0)/60000)}m current=${e.current}`;
     case "auto-update-disabled":

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mslxdff",
-  "version": "0.1.36",
+  "version": "0.1.38",
   "description": "测试项目，请勿使用。",
   "type": "module",
   "bin": {

package/src/routes.js

@@ -643,11 +643,12 @@ async function racePeerCandidates(candidates, ctx) {
       const order = [];
       const total = prepared.length;
       for (const { peer, target } of prepared) {
+        ctx.evt("peer-request", { peer: peer.url, model: target, hops: ctx.hops + 1 });
         const t0 = performance.now();
         forwardToPeer(peer, ctx.body, target, ctx.hops).then((res) => {
           const latencyMs = Math.round(performance.now() - t0);
           const failed = res instanceof Error || res.status >= 400;
-          ctx.evt("peer-forward", { peer: peer.url, model: target, hops: ctx.hops + 1 });
+          ctx.evt("peer-forward", { peer: peer.url, model: target, hops: ctx.hops + 1, latencyMs, ok: !failed });
           if (failed) {
             const status = res instanceof Error ? 502 : res.status;
             ctx.logError(ctx.model,
@@ -815,7 +816,8 @@ const ROUTES = [
             }
             evt("slow-model", { model, elapsedMs: out.totalMs ?? (Date.now() - startedAt), threshold: STALL_TIMEOUT_MS, interrupted: true, detail: out.detail ?? null });
             logCall(model, 200);
-            evt("result", { model, status: out.status, via: "local", timing: upRes._t ?? null, ttfMs: out.ttfMs, totalMs: out.totalMs, interrupted: true, detail: out.detail ?? null });
+            evt("result", { model, status: out.status, via: "local", timing: upRes._t ?? null, ttfMs: out.ttfMs, totalMs: out.totalMs, interrupted: true, detail: out.detail ?? null, fallback, requested, actual: model });
+            evt("client-response", { requested, actual: model, via: "local", fallback, status: out.status, interrupted: true, reqId });
             return;
           }
           const elapsed = Date.now() - startedAt;
@@ -841,7 +843,8 @@ const ROUTES = [
           } else if (scoredSlow && out.detail) {
             // already recorded slow+latency above, still ensure latency EMA is updated for slow case (done)
           }
-          evt("result", { model, status: out.status, via: "local", timing: upRes._t ?? null, ttfMs: out.ttfMs, totalMs: out.totalMs, detail: out.detail ?? null });
+          evt("result", { model, status: out.status, via: "local", timing: upRes._t ?? null, ttfMs: out.ttfMs, totalMs: out.totalMs, detail: out.detail ?? null, fallback, requested, actual: model });
+          evt("client-response", { requested, actual: model, via: "local", fallback, status: out.status, reqId });
           return;
         }
 
@@ -877,7 +880,8 @@ const ROUTES = [
                 await auto.recordOk(win.target, { latencyMs });
               }
             }
-            evt("result", { model: win.target, status: out.status, via: "peer", timing: win.res._t ?? null, ttfMs: out.ttfMs, totalMs: out.totalMs, detail: out.detail ?? null });
+            evt("result", { model: win.target, status: out.status, via: "peer", timing: win.res._t ?? null, ttfMs: out.ttfMs, totalMs: out.totalMs, detail: out.detail ?? null, fallback: peerFallback, requested, actual: win.target });
+            evt("client-response", { requested, actual: win.target, via: "peer", fallback: peerFallback, status: out.status, reqId });
             return;
           }
           evt("peer-race-lose", { reqId, model });
@@ -908,7 +912,8 @@ const ROUTES = [
                   await auto.recordOk(model, { latencyMs });
                 }
               }
-              evt("result", { model, status: out.status, via: "broadband", timing: bb.result._t ?? null, ttfMs: out.ttfMs, totalMs: out.totalMs, detail: out.detail ?? null });
+              evt("result", { model, status: out.status, via: "broadband", timing: bb.result._t ?? null, ttfMs: out.ttfMs, totalMs: out.totalMs, detail: out.detail ?? null, fallback: bbFallback, requested, actual: model });
+              evt("client-response", { requested, actual: model, via: "broadband", fallback: bbFallback, status: out.status, reqId });
               return;
             } else if (bb.result && typeof bb.result.status === "number") {
               // buffered result from local leader enqueue
@@ -935,7 +940,8 @@ const ROUTES = [
                 onDownstreamAbort: () => evt("client-abort", { reqId, model, totalMs: Math.round(performance.now() - perf0), stages: [...stages] }),
               });
               evt("relay-done", { reqId, model, via: "broadband-local", status: out.status, ttfMs: out.ttfMs, totalMs: out.totalMs, aborted: out.aborted, interrupted: out.interrupted ?? false, detail: out.detail ?? null });
-              evt("result", { model, status: out.status, via: "broadband", timing: null, ttfMs: out.ttfMs, totalMs: out.totalMs, detail: out.detail ?? null });
+              evt("result", { model, status: out.status, via: "broadband", timing: null, ttfMs: out.ttfMs, totalMs: out.totalMs, detail: out.detail ?? null, fallback: bbLocalFallback, requested, actual: model });
+              evt("client-response", { requested, actual: model, via: "broadband", fallback: bbLocalFallback, status: out.status, reqId });
               return;
             }
           }