badgr-cli 1.0.34 → 1.0.35

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "badgr-cli",
-  "version": "1.0.34",
+  "version": "1.0.35",
   "description": "Badgr, run or serve GPU workloads from one command",
   "type": "module",
   "bin": {

package/src/api.js

@@ -105,7 +105,7 @@ export function runJob(config, body) {
     apiKey: config.apiKey,
     baseUrl: config.baseUrl,
     body,
-    timeoutMs: 75_000,
+    timeoutMs: 130_000,
   });
 }
 
@@ -115,7 +115,7 @@ export function serveModel(config, body) {
     apiKey: config.apiKey,
     baseUrl: config.baseUrl,
     body,
-    timeoutMs: 75_000,
+    timeoutMs: 130_000,
   });
 }
 

package/src/fallback.js

@@ -24,7 +24,7 @@ export class CapacityError extends Error {
 }
 
 /**
- * Call an API endpoint with automatic tier-2 expansion on NO_CAPACITY_MATCH.
+ * Call an API endpoint with automatic tier-2 expansion when tier-1 fails.
  * Returns the deployment object on success.
  * Throws CapacityError (pre-formatted for display) on unrecoverable failure.
  * Re-throws payment errors (err.isPaymentRequired) for callers to handle.
@@ -44,10 +44,9 @@ export async function callWithFallback(endpoint, callOpts, buildBody, effectiveT
   const cmd   = labels?.cmd   ?? 'badgr run';
   const allowTier2Fallback = opts.allowTier2Fallback !== false; // default true
 
-  // 75s: backend provisioning runs up to 55s and returns PROVISIONING_TIMEOUT;
-  // the extra headroom ensures the CLI always sees the real response.
+  // 130s: backend provisioning runs up to BADGR_PROVISION_TIMEOUT_SECONDS (default 120s).
   async function attempt(body) {
-    return callApi(endpoint, { method: 'POST', ...callOpts, body, timeoutMs: 75_000 });
+    return callApi(endpoint, { method: 'POST', ...callOpts, body, timeoutMs: 130_000 });
   }
 
   // Map server error codes → catalog keys and extract context for templates.
@@ -118,8 +117,8 @@ export async function callWithFallback(endpoint, callOpts, buildBody, effectiveT
     }
   }
 
-  // Tier-2 expansion: try budget-tier providers when primary has no capacity
-  if (firstErr.errorData?.code === 'NO_CAPACITY_MATCH' && effectiveTier !== '2' && allowTier2Fallback) {
+  // Tier-2 expansion: try budget-tier providers whenever tier-1 fails
+  if (effectiveTier !== '2' && allowTier2Fallback) {
     console.log(chalk.dim('\n  Primary capacity unavailable — expanding search...\n'));
     try {
       return await attempt(buildBody('2'));

package/tests/serve-lifecycle.test.js

@@ -270,7 +270,7 @@ describe('invalid model ID or no capacity', () => {
       errorData: { code: 'PROVISIONING_FAILED', debug_error: 'HF_TOKEN missing, repo is private' },
       httpStatus: 503,
     });
-    // tier-1 provisioning fails; tier-2 also fails (or is not tried without NO_CAPACITY_MATCH)
+    // tier-1 provisioning fails; tier-2 is also tried and fails
     api.callApi.mockRejectedValue(provErr);
 
     const errLines = [];