[Store] - Implement partial success handling in BatchOffload #1319
Conversation
…sting - Add SetTestFailurePredicate() virtual method to StorageBackendInterface - Implement test failure injection in StorageBackendAdaptor and OffsetAllocatorStorageBackend - Refactor TestPartialSuccessBehavior helper to use predicate-based injection - Remove repetitive key creation code (use loops) - Add PartialSuccess tests for both Adaptor and OffsetAllocator backends - Update OutOfSpace test to handle partial success semantics (value==0) This enables deterministic testing of partial success behavior without relying on approximate space exhaustion, making tests more reliable.
maheshrbapatu
requested review from
XucSh,
YiXR,
stmatengss and
ykwd
as code owners
Summary of ChangesHello @maheshrbapatu, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request significantly enhances the robustness of the storage backend's batch offload operations. By introducing partial success handling, the system can now gracefully manage scenarios where individual key operations within a batch might fail. Instead of aborting the entire batch, the process continues, ensuring that all possible successful operations are completed and accurately reported. This improvement prevents data inconsistencies and makes the storage system more resilient to transient errors during batch writes. Highlights
🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console. Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request implements partial success handling for BatchOffload in StorageBackendAdaptor and OffsetAllocatorStorageBackend. Instead of failing the entire batch on a single key failure, the operation now continues, and only successfully offloaded keys are reported. This is a good improvement for robustness. The changes are well-tested, with a new test helper TestPartialSuccessBehavior and failure injection mechanism (SetTestFailurePredicate) to deterministically test partial success scenarios.
My feedback includes a suggestion to improve efficiency in OffsetAllocatorStorageBackend by using break instead of continue on space exhaustion errors, and a couple of suggestions to improve code clarity and maintainability in the tests.
There was a problem hiding this comment.
Pull request overview
This PR implements partial success handling in BatchOffload for both StorageBackendAdaptor and OffsetAllocatorStorageBackend. The key improvement is that when some keys fail during batch processing but others succeed, the operation continues processing remaining keys and reports only the successful keys to the master via the complete_handler. This prevents inconsistency where the storage backend has written keys but the master wasn't notified.
- Changed BatchOffload to continue processing on individual key failures instead of aborting the entire batch
- Added test failure injection mechanism to enable deterministic testing of partial success scenarios
- Return the count of successfully processed keys instead of always returning 0
Reviewed changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.
| File | Description |
|---|---|
| mooncake-store/tests/storage_backend_test.cpp | Adds TestPartialSuccessBehavior helper function and two new test cases (AdaptorBatchOffload_PartialSuccess and OffsetAllocatorStorageBackend_PartialSuccess) to verify partial success handling; also updates OutOfSpace test to handle partial success return values |
| mooncake-store/src/storage_backend.cpp | Implements partial success logic in BatchOffload for both StorageBackendAdaptor and OffsetAllocatorStorageBackend by continuing on individual failures and returning the count of successful keys; adds test failure injection support |
| mooncake-store/include/storage_backend.h | Adds SetTestFailurePredicate virtual method to StorageBackendInterface and implements it in StorageBackendAdaptor and OffsetAllocatorStorageBackend with corresponding test_failure_predicate_ member variables |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
|
Thanks for your contribution! The code is well implemented from my perspective. Feel free to check for the review generated by gemini code agent. |
|
Thanks for taking a look at this @zhangzuo21. I have adressed all the review comments now |
|
Codecov Report❌ Patch coverage is 📢 Thoughts on this report? Let us know! |
|
It is a very useful feature. LGTM. Thx. |
4 participants
Description
Implement partial success handling in BatchOffload for both StorageBackendAdaptor and OffsetAllocatorStorageBackend. When some keys fail but others succeed, continue processing and report only successful keys to master via complete_handler. This prevents inconsistency where storage backend has written keys but master wasn't notified.
Type of Change
How Has This Been Tested?
Checklist