Repository navigation
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info
Walkthrough本次变更调整加载失败的记录和传播方式,更新批量及单条插入任务的异常处理与关闭协调,并使加载器根据加载结果返回状态。新增测试覆盖失败日志、并发关闭和 CLI 失败退出。 Changes加载失败处理与关闭流程
Priority: ➖ Normal Estimated code review effort: 4 (Complex) | ~45 minutes Change: Bug fix Sequence Diagram(s)sequenceDiagram
participant HugeGraphLoader
participant TaskManager
participant InsertFuture
participant LoadContext
participant FailWriter
HugeGraphLoader->>TaskManager: 提交插入任务
TaskManager->>InsertFuture: 执行任务
InsertFuture-->>TaskManager: 返回任务异常
TaskManager->>LoadContext: 登记加载失败
TaskManager->>FailWriter: 写入失败记录
HugeGraphLoader->>LoadContext: 关闭上下文并检查失败状态
Merge Risk: ⚪ Minimal · up to This change propagates loader failures to the exit status and tightens the shutdown order. The review found no concrete defect blocking merge. Run the reported pending functional tests and CI before merging. Pre-merge checks |
|
- preserve failed batch records when fallback is disabled - propagate worker and checkpoint errors to the exit status - drain submissions and cancelled tasks before closing resources - flush all failure logs and reject unreliable progress
a74a103 to
a1ab617
Compare
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## master #56 +/- ##
============================================
+ Coverage 46.62% 46.89% +0.27%
- Complexity 4447 4495 +48
============================================
Files 602 602
Lines 29235 29404 +169
Branches 3422 3450 +28
============================================
+ Hits 13630 13789 +159
- Misses 14329 14332 +3
- Partials 1276 1283 +7 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
- avoid the failing Docker Hub authentication endpoint - retain the exact previously successful official image - preserve service settings and all regression assertions
- fetch existing Maven and Java images from official ECR - keep Flink probe and runtime on the same verified image - preserve the distinct Loader and CDC MySQL versions - retain SDK builder overrides and all test assertions
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
- retain the first runtime exception type and message - avoid self suppression during test cleanup - use a deterministic CLI parse failure fixture - verify client closure and rejected progress after failure
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 8f57a84774
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
- retain the existing Docker Hub image sources - remove transient network repairs from the loading fix - preserve all failure and recovery behavior changes
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
- Distinguish configured read limits from error stops - Keep real errors and generic stops authoritative - Avoid failure data for batches skipped at the read limit - Cover queued submissions and stop reason precedence
- drain asynchronous tasks before reporting failures - preserve suppressed errors and interrupt state - retain backup input when restore fails - cover corrupt input and valid empty backups
- use configured graph space in existing schema APIs - preserve legacy routes for empty graph spaces - retain request and HTTP error semantics - cover route escaping and request bodies with race
The ordinary Loader distribution cannot start existing GRAPH or Parquet sources: its selected protobuf runtime lacks the PD discovery classes, and Parquet classes are inherited only through Spark's provided dependencies. The ordinary package now carries the dependencies these existing readers require.
- Incorporate merged Toolchain runtime and consumer fixes - Preserve Loader failure and read limit behavior - Retain all upstream and Loader suite registrations
Purpose of the PR
Loader can report success after parse, insert, or checkpoint failures. With batch fallback disabled, rejected batches can also disappear from both the failure files and the already advanced progress. This change makes unsuccessful loads fail visibly and retains the information needed for replay.
Reported during master integration testing.
Main Changes
--max-read-linesas a successful intentional stop; later real errors still take precedence.The shutdown and recovery changes are delivered together as one independent fix: splitting them would leave unsafe intermediate resource and progress behavior. This does not add per-record acknowledgements or a new replay protocol.
flowchart LR A[Batch fails without fallback] --> B[Retain the batch in failure-data] B --> C[Drain work and flush logs] C --> D{Progress reliable?} D -- Yes --> E[Save parsed progress] D -- No --> F[Keep old progress] E --> G[Return failure] F --> GVerifying these changes
Does this PR potentially affect the following parts?
Documentation Status
Doc - TODODoc - DoneDoc - No Need: the existing progress and failure-data contract is preserved.Native package recovery acceptance passes. The previous upstream head passed all core profiles after one bounded retry of an external SDK download/cache failure; the read-limit correction is covered by focused regressions and is being checked on the updated head. No dependency or image changes were needed.
Synced with upstream master while retaining both upstream and PR regressions. Focused tests and style checks pass; current-head core CI is running.