refactor: extract PathSharder module for shared multi-path selection#2982
Merged
sammshen merged 1 commit intoLMCache:devfrom Apr 14, 2026
Merged
refactor: extract PathSharder module for shared multi-path selection#2982sammshen merged 1 commit intoLMCache:devfrom
sammshen merged 1 commit intoLMCache:devfrom
Conversation
2 tasks
Contributor
There was a problem hiding this comment.
Code Review
This pull request introduces a new PathSharder utility to centralize path-sharding logic for multi-path storage backends, refactoring GdsBackend and LocalDiskBackend to use this shared component. The review identified a potential bug in the _resolve_device_id helper function regarding "cpu" device handling and robustness against malformed device strings, for which a code suggestion was provided.
aea0776 to
fc8c14a
Compare
aaab7b6 to
e3fd684
Compare
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 2 potential issues.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit e3fd684. Configure here.
Collaborator
|
Hello @glimchb, Thanks for the hard work. Could you check bot review? |
Move the duplicated path-sharding logic (CSV parsing, device ID extraction, by_gpu strategy) from LocalDiskBackend and GdsBackend into a new PathSharder class in path_sharder.py. Both backends now delegate to PathSharder, keeping the policy in one place. The _resolve_device_id helper handles cuda:N, bare cuda, and cpu devices gracefully. Generated with [Devin](https://cli.devin.ai/docs) Co-Authored-By: Devin <noreply@cognition.ai> Signed-off-by: Boris Glimcher <Boris.Glimcher@emc.com>
DongDongJu
approved these changes
Apr 13, 2026
Collaborator
DongDongJu
left a comment
There was a problem hiding this comment.
LGTM.
I left one nit question.
|
|
||
| def test_gds_path_sharding_default(self, temp_gds_path, async_loop): | ||
| """Default gds_path_sharding is 'by_gpu'.""" | ||
| """Default gds_path_sharding is 'by_gpu' (backend inits OK).""" |
Collaborator
There was a problem hiding this comment.
What does "backend inits OK" meaning?
ekaynar
pushed a commit
to ekaynar/LMCache
that referenced
this pull request
Apr 15, 2026
…MCache#2982) Move the duplicated path-sharding logic (CSV parsing, device ID extraction, by_gpu strategy) from LocalDiskBackend and GdsBackend into a new PathSharder class in path_sharder.py. Both backends now delegate to PathSharder, keeping the policy in one place. The _resolve_device_id helper handles cuda:N, bare cuda, and cpu devices gracefully. Generated with [Devin](https://cli.devin.ai/docs) Signed-off-by: Boris Glimcher <Boris.Glimcher@emc.com> Co-authored-by: Devin <noreply@cognition.ai>
ekaynar
added a commit
to ekaynar/LMCache
that referenced
this pull request
Apr 15, 2026
- Remove validate_nixl_path method from NixlStorageConfig - Update NixlFilePool to accept PathSharder instance - Update createPool to use PathSharder with buffer_device - Update tests to use PathSharder directly This aligns with PR LMCache#2982 which centralized path sharding logic.
ekaynar
added a commit
to ekaynar/LMCache
that referenced
this pull request
Apr 15, 2026
- Remove validate_nixl_path method from NixlStorageConfig - Update NixlFilePool to accept PathSharder instance - Update createPool to use PathSharder with buffer_device - Update tests to use PathSharder directly This aligns with PR LMCache#2982 which centralized path sharding logic.
ekaynar
added a commit
to ekaynar/LMCache
that referenced
this pull request
Apr 15, 2026
- Remove validate_nixl_path method from NixlStorageConfig - Update NixlFilePool to accept PathSharder instance - Update createPool to use PathSharder with buffer_device - Update tests to use PathSharder directly This aligns with PR LMCache#2982 which centralized path sharding logic. Signed-off-by: Ugur Kaynar <ugur.kaynar@dell.com>
ftian1
pushed a commit
to ftian1/LMCache
that referenced
this pull request
Apr 20, 2026
…MCache#2982) Move the duplicated path-sharding logic (CSV parsing, device ID extraction, by_gpu strategy) from LocalDiskBackend and GdsBackend into a new PathSharder class in path_sharder.py. Both backends now delegate to PathSharder, keeping the policy in one place. The _resolve_device_id helper handles cuda:N, bare cuda, and cpu devices gracefully. Generated with [Devin](https://cli.devin.ai/docs) Signed-off-by: Boris Glimcher <Boris.Glimcher@emc.com> Co-authored-by: Devin <noreply@cognition.ai>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Move the duplicated path-sharding logic (CSV parsing, device ID extraction, by_gpu strategy) from LocalDiskBackend and GdsBackend into a new PathSharder class in path_sharder.py.
Both backends now delegate to PathSharder, keeping the policy in one place. The _resolve_device_id helper handles cuda:N, bare cuda, and cpu devices gracefully.
Generated with Devin
What this PR does / why we need it:
Special notes for your reviewers:
If applicable:
Note
Medium Risk
Touches storage backend initialization and path selection logic for both
LocalDiskBackendandGdsBackend, which could change which directory is used (and how invalid configs fail) in multi-GPU/multi-path deployments.Overview
Refactors multi-path cache directory selection by extracting the shared CSV parsing + by-GPU sharding logic into a new
PathSharderutility and switching bothGdsBackendandLocalDiskBackendto use it.Standardizes device-id resolution (handles
cuda:N, barecuda, andcpu) and changes unsupported sharding configuration failures fromAssertionErrortoValueError, with updated/new unit tests covering selection, directory creation, and edge cases.Reviewed by Cursor Bugbot for commit 1923414. Bugbot is set up for automated code reviews on this repo. Configure here.