You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
perf(ci): rebalance workspace-test shards — shard(1) 39m mean/110m max vs shards 2,3 at 7-8m #5429
Measured 2026-10-16 over last 12 merged PR heads (check-run completedAt-startedAt): workspace-test-shard (1) mean 39m max 110m; (2) 7m; (3) 8m. Critical path is shard 1. Rebalance the nextest partition (or add a 4th/5th shard / use timing-weighted partition) so all shards <=10-18m. Workflow change: non-author quorum sign-off, no waiver. Refs #5427, aprender-pace PACE-6.
a3 (DUTY 0 claim check, 11:08Z): not starting new work. Partial fix on main: #5445 (1f30ea3) moved the quick-tier tree-reader tests to shard 2. The rest of the rebalance is PR #5487 (GPU crates + steps rebalanced via ci/sections.yml + ci/shard-step-costs.tsv; bank PASS 3/3 at 2c42715, cop sign-off posted, seats on file). Its own estimate puts the shard spread at 52 s, but the tsv Σ row is an estimate. Remaining to close this ticket: after #5487 merges, measure shards 1-3 over 12 merged heads (check-run completedAt-startedAt) and confirm all <=10-18m. Left open for the owner of this issue to close.
Mostly done: #5445 merged 2026-10-16T21:06Z. A worker reports a remainder; I did not measure it. Removed ready until the remainder is stated; closing left to the cop/operator.
Activity
ci(workspace-test): spread the quick tier's selected crates across th…
ci(workspace-test): partition the quick tier's tree readers across th…
ci(workspace-test): quick-tier tree-reader tests move to shard 2 (Refs …
noahgift commentedon Oct 18, 2026
a3 (DUTY 0 claim check, 11:08Z): not starting new work. Partial fix on main: #5445 (1f30ea3) moved the quick-tier tree-reader tests to shard 2. The rest of the rebalance is PR #5487 (GPU crates + steps rebalanced via ci/sections.yml + ci/shard-step-costs.tsv; bank PASS 3/3 at 2c42715, cop sign-off posted, seats on file). Its own estimate puts the shard spread at 52 s, but the tsv Σ row is an estimate. Remaining to close this ticket: after #5487 merges, measure shards 1-3 over 12 merged heads (check-run completedAt-startedAt) and confirm all <=10-18m. Left open for the owner of this issue to close.
noahgift commentedon Oct 18, 2026
Mostly done: #5445 merged 2026-10-16T21:06Z. A worker reports a remainder; I did not measure it. Removed
readyuntil the remainder is stated; closing left to the cop/operator.noahgift commentedon Oct 22, 2026
Tracked in #5510 (triage 2026-10-22).