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.

Sign up for GitHub

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

Jump to bottom

[FLINK-35379][Checkpoint] Fix incorrect checkpoint notification handling in file merging #24806

Merged
merged 2 commits into from
May 21, 2024
Merged

[FLINK-35379][Checkpoint] Fix incorrect checkpoint notification handling in file merging #24806

merged 2 commits into from
May 21, 2024

Conversation

Zakelly
Copy link
Contributor

@Zakelly Zakelly commented May 17, 2024 •
edited

What is the purpose of the change

This PR fix several issues with file merging mechanism, see Brief change log.

Brief change log

Fix following issues:

  • Checkpoint notification is not handled by the file merging manager.
  • File manager propagate the discard of one checkpoint incorrectly.
  • File manager cannot recognize the PlaceholderStreamStateHandle
  • Fail to filter EmptySegmentFileStateHandle out during recovery

Verifying this change

UT modified.

Does this pull request potentially affect one of the following parts:

  • Dependencies (does it add or upgrade a dependency): no
  • The public API, i.e., is any changed class annotated with @Public(Evolving): no
  • The serializers: no
  • The runtime per-record code paths (performance sensitive): no
  • Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Kubernetes/Yarn, ZooKeeper: yes
  • The S3 file system connector: no

Documentation

  • Does this pull request introduce a new feature? no
  • If yes, how is the feature documented? not applicable

@Zakelly Zakelly requested a review from fredia May 17, 2024 07:56
@flinkbot
Copy link
Collaborator

flinkbot commented May 17, 2024 •
edited

CI report:

Bot commands The @flinkbot bot supports the following commands:
  • @flinkbot run azure re-run the last Azure build

@Zakelly Zakelly mentioned this pull request May 17, 2024
Merged
@Zakelly
Copy link
Contributor Author

Zakelly commented May 17, 2024

@flinkbot run azure

@Zakelly
Copy link
Contributor Author

Zakelly commented May 20, 2024

@flinkbot run azure

fredia
fredia reviewed May 20, 2024
View reviewed changes
Copy link
Contributor

@fredia fredia left a comment

Choose a reason for hiding this comment

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

Thanks for the PR. I left two comments.

File manager cannot recognize the PlaceholderStreamStateHandle

Theoretically, PlaceholderStreamStateHandle should not be written to the meta file, and File manager should not know it.

...g/apache/flink/runtime/checkpoint/filemerging/SubtaskFileMergingManagerRestoreOperation.java Outdated
.map(handle -> (SegmentFileStateHandle) handle);
System.out.println("Restoring from checkpoint " + checkpointId);
Copy link
Contributor

Choose a reason for hiding this comment

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

LOG.info()?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Debug info attached by mistake. Removed.

...ain/java/org/apache/flink/runtime/checkpoint/filemerging/FileMergingSnapshotManagerBase.java
notifiedCheckpoint.add(checkpointId);
discardCheckpoint(checkpointId);
if (notifiedCheckpoint.size() > NUM_GHOST_CHECKPOINT_IDS) {
notifiedCheckpoint.pollFirst();
Copy link
Contributor

Choose a reason for hiding this comment

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

Should the corresponding checkpoint also be removed from notifiedSubtaskCheckpoint ?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Added in notifyReleaseCheckpoint and tryDiscardCheckpoint

@Zakelly
Copy link
Contributor Author

Zakelly commented May 20, 2024

Thanks for the PR. I left two comments.

File manager cannot recognize the PlaceholderStreamStateHandle

Theoretically, PlaceholderStreamStateHandle should not be written to the meta file, and File manager should not know it.

This statement only applies for the reusePreviousStateHandle, which is used for LogicalFile tracking across consecutive checkpoints.

@Zakelly
Copy link
Contributor Author

Zakelly commented May 20, 2024

@fredia Thanks for your comments! I've addressed them and PTAL.

fredia
fredia approved these changes May 21, 2024
View reviewed changes
Copy link
Contributor

@fredia fredia left a comment

Choose a reason for hiding this comment

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

@Zakelly Thanks for the update, LGTM.

@Zakelly Zakelly merged commit 2bcc308 into apache:master May 21, 2024
@Zakelly Zakelly deleted the f35379 branch May 21, 2024 02:16
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@fredia fredia fredia approved these changes

Assignees
No one assigned
Labels
component=Runtime/Checkpointing
Projects
None yet
Milestone
No milestone
3 participants
@Zakelly @flinkbot @fredia