streaming_shuffle.txt 3.3 KB

123456789101112131415161718192021222324252627282930313233
  1. ubuntu@ip-172-31-52-199:~/data_processing_tests$ python -m ray.experimental.shuffle --num-cpus=32 --num-partitions=200 --partition-size=500e6 --object-store-memory=20e9
  2. Start a new cluster...
  3. 2021-04-15 00:51:14,984 INFO services.py:1269 -- View the Ray dashboard at http://127.0.0.1:8265
  4. Map Progress.: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 200/200 [03:26<00:00, 1.03s/it]
  5. Reduce Progress.: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 200/200 [03:26<00:00, 1.03s/it]
  6. --- Aggregate object store stats across all nodes ---
  7. Plasma memory usage 14401 MiB, 6040 objects, 75.51% full
  8. Spilled 80722 MiB, 33854 objects, avg write throughput 394 MiB/s
  9. Restored 79363 MiB, 33284 objects, avg read throughput 4709 MiB/s
  10. Objects consumed by Ray tasks: 95377 MiB.
  11. Shuffled 95367 MiB in 208.47784852981567 seconds
  12. ubuntu@ip-172-31-52-199:~/data_processing_tests$ python -m ray.experimental.shuffle --num-cpus=8 --num-partitions=200 --partition-size=500e6 --object-store-memory=5e9 --num-nodes=4
  13. Emulating a cluster...
  14. Num nodes: 4
  15. Num CPU per node: 8
  16. Object store memory per node: 5000000000.0
  17. 2021-04-15 01:14:16,310 INFO worker.py:654 -- Connecting to existing Ray cluster at address: 172.31.52.199:6379
  18. Map Progress.: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 200/200 [04:08<00:00, 1.24s/it]
  19. Reduce Progress.: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 200/200 [04:08<00:00, 1.24s/it]
  20. --- Aggregate object store stats across all nodes ---
  21. Plasma memory usage 3459 MiB, 1451 objects, 18.14% full
  22. Spilled 91357 MiB, 38314 objects, avg write throughput 371 MiB/s
  23. Restored 87405 MiB, 36657 objects, avg read throughput 4219 MiB/s
  24. Objects consumed by Ray tasks: 95377 MiB.
  25. Shuffled 95367 MiB in 250.11851406097412 seconds
  26. 2021-04-15 01:18:34,141 ERROR worker.py:928 -- print_logs: Connection closed by server.
  27. 2021-04-15 01:18:34,143 ERROR worker.py:1087 -- listen_error_messages_raylet: Connection closed by server.
  28. 2021-04-15 01:18:34,143 ERROR import_thread.py:88 -- ImportThread: Connection closed by server.