site update Wed Nov 5 00:28:14 PST 2025

Author: akirillov

declarative-kubernetes-cluster-emulation-with-kemu/index.html

@@ -139,7 +139,7 @@
     "keywords": ["kubernetes","emulation","kind","kwok","kemu"],
 
     "mainEntityOfPage": "true",
-    "wordCount": "2568"
+    "wordCount": "2577"
   }
   </script>
 
@@ -456,7 +456,7 @@ <h1 class="mb-8 mt-0 text-4xl font-extrabold text-neutral-900 dark:text-neutral"
 <p>Maximizing infrastructure efficiency through optimized scheduling and high-density bin packing is
 critical for achieving high workload throughput and resource utilization. Introducing scheduler
 modifications at production scale is risky: configuration errors can render workloads unschedulable,
-causing multi-day delays for dozens or hundreds of AI practitioners, idling capacity, wasted on-demand
+causing multi-day delays for dozens or hundreds of ML engineers, idling capacity, wasted on-demand
 resources, and delayed delivery. It is imperative that scheduler modifications with a big blast
 radius are tested and verified in a safe environment before shipping to production.</p>
 <p>This problem is not new, and virtual/emulated clusters for scheduling experimentation and workload
@@ -1103,9 +1103,9 @@ <h3 id="suboptimal-scheduling-example" class="relative group">Suboptimal schedul
 <p>From the dashboard, we can see that while all jobs are in the <code>Running</code> state, there are pods
 in the <code>Pending</code> state while unallocated capacity remains available.</p>
 <p>What we&rsquo;re observing is partial admission - a common scheduling problem in Kubernetes
-clusters. Pods created by jobs are not scheduled in batch but rather one-by-one
-resulting in fragmentation and inability to schedule all pods even if there is
-cluster capacity available.</p>
+clusters. The Kubernetes Scheduler processes pods individually without job-level awareness.
+Pods created by jobs are not scheduled in batch but rather one-by-one resulting in
+fragmentation and inability to schedule all pods even if there is cluster capacity available.</p>
 <p>For distributed training workloads that require all workers to start simultaneously
 (all-or-nothing semantics), this partial admission creates a deadlock situation.
 This simple experiment reveals a scheduling issue that would be expensive to discover