8000 [Core] Error in external storage writing for object spilling · Issue #33913 · ray-project/ray · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content
[Core] Error in external storage writing for object spilling #33913
Open
@joanna-yoo

Description

@joanna-yoo

What happened + What you expected to happen

I'm running ray serve with k8, and I set up the object spilling config as follows:

containers:
  - name: ray-worker
       env:
          - name: "RAY_object_spilling_config"
            value: "{\"type\":\"smart_open\",\"params\":{\"uri\":\"gs://some/bucket/path\"}}"

But this is the error that I got for all of the writes.

Traceback (most recent call last):
  File "python/ray/_raylet.pyx", line 1249, in ray._raylet.spill_objects_handler
  File "python/ray/_raylet.pyx", line 1252, in ray._raylet.spill_objects_handler
  File "//.pyenv/versions/3.10.4/lib/python3.10/site-packages/ray/_private/external_storage.py", line 668, in spill_objects
    return _external_storage.spill_objects(object_refs, owner_addresses)
  File "//.pyenv/versions/3.10.4/lib/python3.10/site-packages/ray/_private/external_storage.py", line 541, in spill_objects
    return self._write_multiple_objects(
  File "//.pyenv/versions/3.10.4/lib/python3.10/site-packages/ray/_private/external_storage.py", line 150, in _write_multiple_objects
    assert written_bytes == payload_len

Under which circumstances would this be false?

Versions / Dependencies

Docker base image: rayproject/ray:2.2.0-py310-gpu
ray[serve]==2.2.0

Reproduction script

I cannot reproduce it in dev setup unfortunately. : ( This works fine locally.

import ray
import smart_open
from ray._private import external_storage

buf = b'0' * 1024 * 1024
object_ref = ray.put(buf)
f = smart_open.open('gs://some/path/test.pkl'), 'wb')
e = external_storage.ExternalStorageSmartOpenImpl('gs://some/path')
e._write_multiple_objects(f, [object_ref], [b'test_address'], 'gs://some/path')

Issue Severity

Medium: It is a significant difficulty but I can work around it.

Metadata

Metadata

Assignees

No one assigned

    Labels

    P3Issue moderate in impact or severitybugSomething that is supposed to be working; but isn'tcoreIssues that should be addressed in Ray Corepending-cleanupThis issue is pending cleanup. It will be removed in 2 weeks after being assigned.

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions

      0