Repository navigation
Benchmark regression: open (main) #573
Description
Bencher flagged these benchmarks in the nightly run (open, main) on two runs in a row:
open/wal_replay/100k
Branch: main
Run: https://github.com/project-minigraf/minigraf/actions/runs/38373154795
Previous run: https://github.com/project-minigraf/minigraf/actions/runs/38225035471
Please review the Bencher dashboard to confirm the regression.
Activity
adityamukho commentedon Oct 13, 2026
Not a code regression. This is the same step as #561, re-alerting.
open/wal_replay/100kon main (Bencher, ubuntu-latest):The step comes from #540 (backport of #476): before it, the
usize::MAXclose-time drop still saved the.graph, so this fixture replayed 0 WAL entries. It now replays all 100k, which is the work the benchmark is meant to measure. The latency has been flat at ~3 s since the step. Commits between the two alert runs (72d847e..d604cc4) are CI, a unit test and a fuzz corpus only; 1k and 10k are unchanged.#568 (merged today) limits the comparison to the last 30 runs, so a lasting step alerts for a few more runs and then becomes the baseline. Further re-fires of this benchmark at ~3 s can be closed the same way.
🤖 Generated with Claude Code