Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Replaceing KV actor by Ray API #136

Merged
merged 5 commits into from
Jan 9, 2024
Merged

Replaceing KV actor by Ray API #136

merged 5 commits into from
Jan 9, 2024

Conversation

NKcqx
Copy link
Collaborator

@NKcqx NKcqx commented Jun 26, 2023

Resolves #135

@NKcqx NKcqx added the enhancement New feature or request label Jun 26, 2023
@NKcqx NKcqx requested review from fengsp and a team June 26, 2023 03:08
Copy link
Collaborator

@jovany-wang jovany-wang left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

As discussed offline, we should enable tests in client mode.

return ray.get(o)
# Note(NKcqx): internval_kv is always initiated after `ray.init`,
# calling this is equal to directly return "True"
return self._client_api_stub._internal_kv_initialized()
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

So we should mark this method as deprecated?


def reset(self):
o = self._internal_kv_actor.reset.remote()
return ray.get(o)
# Note(NKcqx): No `gcs_client` is instantiated for kv, and the 'initialized'
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Mark as deprecated.

Signed-off-by: NKcqx <892670992@qq.com>
@NKcqx
Copy link
Collaborator Author

NKcqx commented Jun 29, 2023

Unfortunately, Ray's client mode internal_kv accessing has a BUG that causes the call of internal_kv_del takes no effect. This PR can't be merged untill this issue is fixed.

@NKcqx
Copy link
Collaborator Author

NKcqx commented Jul 4, 2023

It turns out to be a dumb BUG in Ray, fix PR:
https://github.com/ray-project/ray/pull/37031/files

@jovany-wang
Copy link
Collaborator

If the current ray doesn't support it, how can we handle to get compatible with all versions?

@jovany-wang
Copy link
Collaborator

Do we need go with this PR? @ray-project/rayfed

Signed-off-by: NKcqx <892670992@qq.com>

Conflicts:
	fed/_private/compatible_utils.py
	tests/test_internal_kv.py
Signed-off-by: NKcqx <892670992@qq.com>
Signed-off-by: NKcqx <892670992@qq.com>
Copy link
Collaborator

@jovany-wang jovany-wang left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@jovany-wang jovany-wang merged commit f21251e into main Jan 9, 2024
13 checks passed
@jovany-wang jovany-wang deleted the replace_kv_actor branch January 9, 2024 03:13
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request pending-discussing
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[Multi-Job] The KV Actor increase management burdens of Multi-Job
2 participants