{
  "$schema": "https://themachinepress.com/schemas/story-v1.schema.json",
  "schema_version": "1.0.0",
  "document_type": "machine_press_story",
  "story": {
    "story_id": "mp-2026-08-21-001",
    "source_story_id": "tmp-lead-lazy-pod-deferred-failure",
    "edition_id": "mp-2026-08-21-morning-0043",
    "edition_url": "https://themachinepress.com/edition/2026-08-21",
    "position": 1,
    "story_type": "lead",
    "section": "infrastructure",
    "editorial_classification": "editorial",
    "headline": "The Pod Was Ready. The Files Were Already Failing",
    "slug": "the-pod-was-ready-the-files-were-already-failing",
    "dek": "Lazy model-image pulling cut cold startup to about 17 seconds, then hid cache exhaustion from every visible health check for more than three minutes.",
    "summary": "Lazy model-image pulling cut cold startup to about 17 seconds, then hid cache exhaustion from every visible health check for more than three minutes.",
    "body_text": "Lazy container-image pulling promises to mount an AI model immediately and fetch its contents only when read. In measurements spanning artifacts from 2 to 140 gigabytes, eStargz and AWS SOCI made cold time-to-first-prediction nearly independent of image size: 16.9 to 17.6 seconds, versus 24.5 to 573 seconds for eager pulls. But the cost moved rather than disappeared. Reading a full 14-gigabyte model through the lazy mount took 105.3 seconds, slower than the 72.4-second eager pull it replaced.\n\nThe sharper result was operational. Sustained legitimate reads exhausted the snapshotter's finite node cache, and already-running pods began losing access to model files. At the first failure stage, one instrumented pod passed every Kubernetes-visible and application-level check for 196 seconds while the snapshotter logged real errors. Under heavier pressure, 67 to 94 percent of model files failed; restarting the daemon left stale file handles inside a pod still marked Running. The measurements expose a monitoring and capacity-planning hazard, not a claim that every lazy-pulling deployment will fail.",
    "why_it_matters": "Lazy model-image pulling cut cold startup to about 17 seconds, then hid cache exhaustion from every visible health check for more than three minutes.",
    "limitations": [
      "Lazy container-image pulling promises to mount an AI model immediately and fetch its contents only when read.",
      "The measurements expose a monitoring and capacity-planning hazard, not a claim that every lazy-pulling deployment will fail."
    ],
    "importance": 10,
    "canonical_url": "https://themachinepress.com/story/mp-2026-08-21-001/the-pod-was-ready-the-files-were-already-failing",
    "json_url": "https://themachinepress.com/story/mp-2026-08-21-001.json",
    "first_published_at": "2026-08-21T09:00:00.000-04:00",
    "modified_at": "2026-08-21T09:00:00.000-04:00",
    "content_status": "new",
    "is_carryover": false,
    "carryover_reason": null,
    "key_claims": [
      {
        "claim_id": "claim-mp-2026-08-21-001-001",
        "text": "Lazy model-image pulling cut cold startup to about 17 seconds, then hid cache exhaustion from every visible health check for more than three minutes.",
        "source_ids": [
          "source-2026-08-21-001"
        ],
        "qualification": "Lazy container-image pulling promises to mount an AI model immediately and fetch its contents only when read."
      }
    ],
    "source_ids": [
      "source-2026-08-21-001"
    ],
    "tags": [
      "Kubernetes",
      "model serving",
      "lazy image pulling",
      "cache exhaustion"
    ],
    "image_url": "https://themachinepress.com/issues/2026-08-21/lead-lazy-pod-deferred-failure.png",
    "corrections": []
  },
  "sources": [
    {
      "source_id": "source-2026-08-21-001",
      "title": "arXiv preprint 2608.19412",
      "publisher": "arXiv",
      "url": "https://arxiv.org/abs/2608.19412",
      "canonical_url": "https://arxiv.org/abs/2608.19412",
      "source_type": "primary_research",
      "is_primary_source": true,
      "published_at": "2026-08-18T20:00:00.000-04:00",
      "accessed_at": "2026-08-21T08:22:06.000-04:00",
      "supports_claim_ids": [
        "claim-mp-2026-08-21-001-001"
      ]
    }
  ],
  "corrections": [],
  "publisher": {
    "name": "The Machine Press",
    "url": "https://themachinepress.com",
    "description": "A daily newspaper for the age of artificial intelligence."
  },
  "cite_this_report": {
    "title": "The Pod Was Ready. The Files Were Already Failing",
    "publisher": "The Machine Press",
    "published_at": "2026-08-21T09:00:00.000-04:00",
    "canonical_url": "https://themachinepress.com/story/mp-2026-08-21-001/the-pod-was-ready-the-files-were-already-failing"
  }
}