From fae358dc092df2fa3d514be009d3a72aa8745d92 Mon Sep 17 00:00:00 2001
From: Intern Dev <dev@wukongai.io>
Date: Mon, 13 Apr 2026 00:48:59 -0400
Subject: [PATCH] fix(nim): bound curl health probes with --max-time

The NIM health probe loop and nimStatusByName() use `curl -sf` without a
timeout. On a hung or silently-dropping NIM endpoint the curl call can
block indefinitely, stalling onboard/waitForNimHealth and CLI status.

Add `--max-time 5` to both probe calls so each attempt times out cleanly
and the outer retry loop can make progress.
---
 src/lib/nim.ts | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/src/lib/nim.ts b/src/lib/nim.ts
index d0118ff393a..5e6b9d9200d 100644
--- a/src/lib/nim.ts
+++ b/src/lib/nim.ts
@@ -209,9 +209,10 @@ export function waitForNimHealth(port = 8000, timeout = 300): boolean {
 
   while ((Date.now() - start) / 1000 < timeout) {
     try {
-      const result = runCapture(`curl -sf http://localhost:${hostPort}/v1/models`, {
-        ignoreError: true,
-      });
+      const result = runCapture(
+        `curl -sf --max-time 5 http://localhost:${hostPort}/v1/models`,
+        { ignoreError: true },
+      );
       if (result) {
         console.log("  NIM is healthy.");
         return true;
@@ -263,7 +264,7 @@ export function nimStatusByName(name: string, port?: number): NimStatus {
         resolvedHostPort = m ? Number(m[1]) : 8000;
       }
       const health = runCapture(
-        `curl -sf http://localhost:${resolvedHostPort}/v1/models 2>/dev/null`,
+        `curl -sf --max-time 5 http://localhost:${resolvedHostPort}/v1/models 2>/dev/null`,
         { ignoreError: true },
       );
       healthy = !!health;