-
Notifications
You must be signed in to change notification settings - Fork 13k
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.
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
[FLINK-18403][checkpointing] Ensure that unaligned checkpointing is only activated for EXACTLY_ONCE (1.11). #12741
Conversation
…ithout parameters. CheckpointConfig#enableUnalignedCheckpoints(boolean) makes it explicit and also is more future-proof. When unaligned checkpoints become the default, this method will be mostly useless and we would need to add a #disableUnalignedCheckpoints() for consistency.
…nly activated for EXACTLY_ONCE.
Thanks a lot for your contribution to the Apache Flink project. I'm the @flinkbot. I help the community Automated ChecksLast check on commit 4a76406 (Mon Jun 22 11:18:18 UTC 2020) Warnings:
Mention the bot in a comment to re-run the automated checks. Review Progress
Please see the Pull Request Review Guide for a full explanation of the review process. The Bot is tracking the review progress through labels. Labels are applied according to the order of the review items. For consensus, approval by a Flink committer of PMC member is required Bot commandsThe @flinkbot bot supports the following commands:
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM.
nit: the Javadoc for enableUnalignedCheckpoints(boolean)
contains a typo: reduce
should be reduces
.
Did you also consider to throw instead of only logging a warning? I can see the argument go both ways (and I'm fine with both), but configuring unaligned checkpoints could also be read as a strong indication that I actually want exactly once rather. In that case it could be surprising to see checkpointing be executed in at least once mode.
Yes, we considered it @uce. We were actually swinging both directions and the deciding factor was that there are quite a bit of individual tests that are manually setting at least once mode. With hard crash it would add quite a bit of effort to modify all of those places to manually disable unaligned checkpoints (we are enabling unaligned checkpoints by default in our tests). We could also choose to refine this behaviour in the future. |
(I've fixed the typo as separate hotfix commit) |
(Unchanged backport of #12739)
What is the purpose of the change
Currently, unaligned checkpoint config option can be used together with AT_LEAST_ONCE (or disabled checkpointing), which may lead to unexpected behavior.
This PR ensures that unaligned checkpointing is only activated for EXACTLY_ONCE.
Brief change log
Verifying this change
Added two unit tests.
Does this pull request potentially affect one of the following parts:
@Public(Evolving)
: (yes / no)Documentation