engine: add input grace period and check pending chunks on shutdown #9952
Conversation
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
singholt
force-pushed
the
engine-grace-input
branch
from
76c983b to
fb13aeb
Compare
singholt
changed the title
singholt
requested review from
edsiper,
leonardo-albertovich,
fujimotos and
koleini
as code owners
singholt
changed the title
|
Hi @edsiper / @leonardo-albertovich could you please review this PR and provide your feedback? thanks! |
edsiper
requested changes
Apr 29, 2025
singholt
force-pushed
the
engine-grace-input
branch
from
fb13aeb to
193394f
Compare
singholt
force-pushed
the
engine-grace-input
branch
from
193394f to
80388ef
Compare
singholt
force-pushed
the
engine-grace-input
branch
from
80388ef to
ccbcbf8
Compare
1. Input grace period Currently, Fluent Bit pauses all inputs 1 second after SIGTERM. This change creates an input grace period, which by default is one half the total Grace setting. This means that half way through the grace period Fluent Bit stops accepting any new logs and only sends logs pending in the buffers. 2. Check pending chunks on shutdown Previously the engine shutdown immediately if there were no pending tasks. A task is created from a chunk in the buffer. If there is a new chunk, but no task yet, the engine should keep running until the task is created and completed. This change makes the engine wait on shutdown for all pending chunks until the max grace period has expired. Signed-off-by: Wesley Pettit <[email protected]> Co-authored-by: Anuj Singh <[email protected]>
Signed-off-by: Wesley Pettit <[email protected]>
…ault: off) Signed-off-by: Anuj Singh <[email protected]>
singholt
force-pushed
the
engine-grace-input
branch
from
ccbcbf8 to
77d61d1
Compare
singholt
added a commit
to singholt/fluent-bit-docs
that referenced
this pull request
May 19, 2025
See corresponding PR: fluent/fluent-bit#9952 Signed-off-by: Anuj Singh <[email protected]>
singholt
added a commit
to singholt/fluent-bit-docs
that referenced
this pull request
May 19, 2025
See corresponding PR: fluent/fluent-bit#9952 Signed-off-by: Anuj Singh <[email protected]>
|
Docs PR: fluent/fluent-bit-docs#1667 |
|
I'll review this PR tomorrow. |
Signed-off-by: Anuj Singh <[email protected]> Co-authored-by: Wesley Pettit <[email protected]>
singholt
force-pushed
the
engine-grace-input
branch
from
77d61d1 to
38850e2
Compare
|
Are you done making changes @singholt? I want to review this today but only if it's the final code. |
Yes, just fixed the compilation error CI caught! Its ready for your review. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
What does this PR do?
This PR makes the following changes:
Currently, Fluent Bit pauses all inputs 1 second after
SIGTERM. This PR creates an input grace period, which by default is half the total "Grace" setting. This means that half way through the grace period Fluent Bit stops accepting any new logs and only sends logs pending in the buffers.Previously the engine shutdown immediately if there were no pending tasks. A task is created from a chunk in the buffer. If there is a new chunk, but no task yet, the engine should keep running until the task is created and completed. This change makes the engine wait on shutdown for all pending chunks until the max grace period has expired.
What use-case does this PR aim to solve?
In production environments with high-throughput logging, applications can generate significant volumes of logs even during the shutdown phase. Container orchestration services, such as Amazon ECS, provide containers with a (configurable) graceful shutdown period (default is 30 seconds in ECS) to properly terminate their operations. However, the current implementation may lead to dropped logs during this shutdown process, as it immediately stops accepting inputs after
SIGTERMand may not process all buffered data.By introducing an input grace period and improving the pending chunk verification, Fluent Bit can now better utilize the provided shutdown window - continuing to accept critical logs for a portion of the grace period while ensuring all buffered data is properly processed and delivered to their destinations. This results in more meaningful use of the shutdown time rather than simply discarding unprocessed input.
These improvements are also valuable when applications perform controlled shutdowns due to conditions like OOM or health check failures - capturing crucial diagnostic logs during the application's final moments.
Fluent Bit is licensed under Apache 2.0, by submitting this pull request I understand that this code will be released under the terms of that license.