[
  {
    "run_id": "run-1001",
    "cluster": "cust-a-seoul",
    "framework": "pytorch",
    "started_at": "2026-08-18T02:11:00Z",
    "duration_sec": 412,
    "exit_code": 1,
    "gpu_type_raw": "A100-80G",
    "log_tail": "RuntimeError: CUDA out of memory. Tried to allocate 2.20 GiB (GPU 0; 79.15 GiB total capacity; 76.02 GiB already allocated)"
  },
  {
    "run_id": "run-1002",
    "cluster": "cust-a-seoul",
    "framework": "pytorch",
    "started_at": "2026-08-18T03:40:00Z",
    "duration_sec": 1810,
    "exit_code": 1,
    "gpu_type_raw": "A100 80GB",
    "log_tail": "torch.distributed.DistBackendError: NCCL error: unhandled system error, watchdog caught collective operation timeout: WorkNCCL(SeqNum=1204, OpType=ALLREDUCE)"
  },
  {
    "run_id": "run-1003",
    "cluster": "cust-b-us",
    "framework": "vllm",
    "started_at": "2026-08-18T05:02:00Z",
    "duration_sec": 8,
    "exit_code": 127,
    "gpu_type_raw": "H100",
    "log_tail": "Failed to pull image \"registry.internal/vllm-serve:0.9.2\": rpc error: code = NotFound desc = manifest unknown"
  },
  {
    "run_id": "run-1004",
    "cluster": "cust-b-us",
    "framework": "vllm",
    "started_at": "2026-08-18T06:20:00Z",
    "duration_sec": null,
    "exit_code": null,
    "gpu_type_raw": "H100",
    "log_tail": "admission webhook denied the request: exceeded quota: gpu-quota, requested: nvidia.com/gpu=8, used: nvidia.com/gpu=8, limited: nvidia.com/gpu=8"
  },
  {
    "run_id": "run-1005",
    "cluster": "cust-a-seoul",
    "framework": "pytorch",
    "started_at": "2026-08-18T09:15:00Z",
    "duration_sec": 33,
    "exit_code": 1,
    "gpu_type_raw": "A100-80G",
    "log_tail": "FileNotFoundError: [Errno 2] No such file or directory: '/mnt/ckpt/stage2/last.ckpt'"
  },
  {
    "run_id": "run-1006",
    "cluster": "cust-c-eu",
    "framework": "sglang",
    "started_at": "2026-08-19T01:05:00Z",
    "duration_sec": 1204,
    "exit_code": 0,
    "gpu_type_raw": "L40S",
    "log_tail": "INFO server started on 0.0.0.0:30000, 128 running requests, avg ttft 182ms"
  },
  {
    "run_id": "run-1007",
    "cluster": "cust-c-eu",
    "framework": "sglang",
    "started_at": "2026-08-19T02:31:00Z",
    "duration_sec": 96,
    "exit_code": 1,
    "gpu_type_raw": "L40S",
    "log_tail": "RuntimeError: CUDA out of memory. Tried to allocate 512.00 MiB (GPU 3; 44.32 GiB total capacity; 43.11 GiB already allocated)"
  },
  {
    "run_id": "run-1008",
    "cluster": "cust-b-us",
    "framework": "pytorch",
    "started_at": "2026-08-19T04:44:00Z",
    "duration_sec": 2402,
    "exit_code": 1,
    "gpu_type_raw": "h100",
    "log_tail": "torch.distributed.DistBackendError: NCCL error: remote process exited or there was a network error"
  },
  {
    "run_id": "run-1009",
    "cluster": "cust-a-seoul",
    "framework": "vllm",
    "started_at": "2026-08-19T07:12:00Z",
    "duration_sec": 5,
    "exit_code": 127,
    "gpu_type_raw": "A100-80G",
    "log_tail": "Failed to pull image \"registry.internal/vllm-serve:latest\": rpc error: code = Unauthorized desc = authentication required"
  },
  {
    "run_id": "run-1010",
    "cluster": "cust-c-eu",
    "framework": "pytorch",
    "started_at": "2026-08-19T11:30:00Z",
    "duration_sec": 7,
    "exit_code": 1,
    "gpu_type_raw": "L40S",
    "log_tail": "FileNotFoundError: [Errno 2] No such file or directory: '/mnt/data/tokenizer.json'"
  },
  {
    "run_id": "run-1011",
    "cluster": "cust-b-us",
    "framework": "sglang",
    "started_at": "2026-08-20T00:18:00Z",
    "duration_sec": 640,
    "exit_code": 0,
    "gpu_type_raw": "H100",
    "log_tail": "INFO server started on 0.0.0.0:30000, 64 running requests, avg ttft 210ms"
  },
  {
    "run_id": "run-1012",
    "cluster": "cust-a-seoul",
    "framework": "pytorch",
    "started_at": "2026-08-20T03:03:00Z",
    "duration_sec": 188,
    "exit_code": 137,
    "gpu_type_raw": "A100-80G",
    "log_tail": "Killed by signal 9 (SIGKILL). container exceeded memory limit 220Gi"
  },
  {
    "run_id": "run-1013",
    "cluster": "cust-c-eu",
    "framework": "vllm",
    "started_at": "2026-08-20T05:50:00Z",
    "duration_sec": null,
    "exit_code": null,
    "gpu_type_raw": "L40S",
    "log_tail": "admission webhook denied the request: exceeded quota: gpu-quota, requested: nvidia.com/gpu=4, used: nvidia.com/gpu=6, limited: nvidia.com/gpu=8"
  },
  {
    "run_id": "run-1014",
    "cluster": "cust-b-us",
    "framework": "pytorch",
    "started_at": "2026-08-20T08:26:00Z",
    "duration_sec": 55,
    "exit_code": 1,
    "gpu_type_raw": "H100 80GB",
    "log_tail": "ValueError: Tokenizer class LlamaTokenizerFast does not exist or is not currently imported"
  }
]