Remove no-op bias from mutation load generation - #24
Merged
Merged
Conversation
bweis
added this pull request to stack #22
September 19, 2026 07:26
bweis
force-pushed
the
bench/mutation-targets
branch
from
September 19, 2026 07:30
fc6901f to
73e6cc3
Compare
bweis
removed this pull request from stack #22
September 19, 2026 07:30
bweis
added this pull request to stack #26
September 19, 2026 07:30
bweis
marked this pull request as ready for review
September 19, 2026 07:37
bweis
force-pushed
the
bench/mutation-targets
branch
from
September 19, 2026 07:38
73e6cc3 to
ab93a87
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
High offered write rates previously targeted not-yet-created IDs and spawned a shell for every no-op, obscuring actual useful throughput. Writers now select and lock a live row in the current key range, wrap gaps, and assert exactly one affected row atomically. Failed/exhausted selections invalidate the run; completed transactions cannot silently be no-ops.
The protocol remains a key-space sample rather than uniform row sampling. Selection cost is included in timings, and the new protocol is explicitly distinguished from historical results. No runtime index code changed.
Validation: 88 benchmark tests, source-header checks, a native sparse/locked-target and rollback regression, and a PG18 30-second two-writer smoke at 8,000 offered mutations/sec passed. All 239,432 completed mutations affected a row; first/last-decile writer lag p95 was 0.44/0.45 ms. This is a short synthetic protocol check, not a production capacity claim. Native regression is added to all four CI jobs.
Stacked on #23.