perf: Reduce online serving hot-path overhead from metrics and audit logging - #6854
Merged
ntkathole merged 1 commit intoSep 22, 2026
Merged
Conversation
…logging Signed-off-by: ntkathole <nikhilkathole2683@gmail.com>
ntkathole
force-pushed
the
perf/online-serving-hot-path-overhead
branch
from
September 22, 2026 05:06
af09314 to
63d9516
Compare
|
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #6854 +/- ##
=======================================
Coverage 47.62% 47.62%
=======================================
Files 422 422
Lines 52352 52358 +6
Branches 7596 7599 +3
=======================================
+ Hits 24931 24936 +5
Misses 25646 25646
- Partials 1775 1776 +1
... and 1 file with indirect coverage changes Continue to review full report in Codecov by Harness.
🚀 New features to boost your workflow:
|
jyejare
approved these changes
Sep 22, 2026
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
3 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 this PR does / why we need it:
Reduces per-request overhead on the
/get-online-featuresendpoint that wasintroduced between v0.61.0 and v0.65.0. Users upgrading across this range
observed p99 latency increase from ~80 ms to ~200 ms with the same
worker/replica configuration.
Two root causes are addressed:
1. Count PRESENT statuses incrementally (
utils.py)track_feature_statuses()(added in #6280) scans the entirefeat_statusesmatrix after it is already built:
For a request with 25 features × 100 entities × 3 feature views, this
re-scans 7,500 status values. This PR counts PRESENT values during the
existing assignment loop, making the counting O(1) additional work:
2. Skip audit overhead when
audit_loggingis disabled (feature_server.py)_emit_online_audit()(added in #6340) runs in thefinallyblock of everyonline request. Even with
audit_logging: false(the default), every requeststill:
time.monotonic()twice for audit timingget_security_manager()to resolve requestor identity_parse_feature_info()to re-parse the feature listThis PR adds an early-exit guard on
feast_metrics._config.audit_loggingsothat when audit logging is disabled, none of this work is performed.