fix: restore all checkpoint state data during resume (#4519)#4526
Open
4yDX3906 wants to merge 1 commit intoalibaba:mainfrom
Open
fix: restore all checkpoint state data during resume (#4519)#45264yDX3906 wants to merge 1 commit intoalibaba:mainfrom
4yDX3906 wants to merge 1 commit intoalibaba:mainfrom
Conversation
The OverAllState.input() method was filtering out keys that didn't have a registered KeyStrategy, causing checkpoint state data to be lost when resuming from suspension. This fix makes input() consistent with updateState() by using the default REPLACE strategy for unregistered keys, ensuring all checkpoint state data is properly restored on resume. Closes alibaba#4519 🤖 Generated with [Qoder][https://qoder.com]
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Fixes #4519.
Problem
When a node stores data in
OverAllStateand then the graph suspends (interrupt), the subsequent node cannot access that data afterresume(). This is becauseOverAllState.input()— called during resume to restore checkpoint state — only processes keys that have a registeredKeyStrategy, silently dropping all other keys.In contrast, during normal execution,
updateState()uses a defaultREPLACEstrategy for unregistered keys, so nodes can write to arbitrary keys successfully. This inconsistency causes checkpoint state data loss on resume.Solution
Modified
OverAllState.input()to use the defaultREPLACEstrategy for keys without a registeredKeyStrategy, making it consistent withupdateState(). This ensures all checkpoint state data is properly restored when resuming from suspension.Changes
OverAllState.java: Changedinput()to process all keys (not just those with registered strategies), usingKeyStrategy.REPLACEas the default fallbackInterruptionTest.java: AddedresumeShouldPreserveCheckpointStatetest that reproduces the exact scenario from the issueTesting
resumeShouldPreserveCheckpointStatepasses — verifies Node B can read Node A's state data after resumeInterruptionTestcases pass🤖 Generated with [Qoder][https://qoder.com]