mirror of
https://github.com/stablyai/orca.git
synced 2026-09-30 16:02:56 +00:00
Neil
f8f656ca19
A concurrency slot is charged per job, not per core, and the account's cap is the scarce resource: standard runner minutes are free and unlimited on a public repository. Two paths spent slots that bought nothing. The unit matrix ran eight fixed shards averaging 6.5 minutes each, 3384 job-slots a day and 68% of all slot demand, while the arm pool queued 10.5 minutes at p95 — the queue was the oversharding. Five shards run the same work in ~10.5 minutes each for three fewer slots per run. Bun profile persistence escalated to all six platforms on `config/`, `resources/` and `.github/` wholesale, which took 36.5% of the last 1100 commits through the full matrix where a platform-flavoured predicate takes 19%. A pull request now qualifies one platform unless the change is platform- flavoured, and the push to main re-qualifies all six, so an unescalated miss surfaces minutes after merge rather than at the next cron. Missing changed-file evidence and an unavailable dependency graph still fail closed to all six.
69 lines
2.7 KiB
JavaScript
69 lines
2.7 KiB
JavaScript
import { unitConsumers } from './ci-unit-dependency-graph.mjs'
|
|
import { balanceFiles } from './ci-shard-assignment.mjs'
|
|
|
|
export function selectUnitFiles(files, changed, graph) {
|
|
const full = (reason) => ({ files, reason, full: true })
|
|
if (!changed.length) {
|
|
return full('Missing changed-path evidence')
|
|
}
|
|
if (changed.some((file) => !file.startsWith('src/') || !graph.files.has(file))) {
|
|
return full('Global, deleted, renamed or unknown input')
|
|
}
|
|
for (const file of changed) {
|
|
const consumers = unitConsumers([file], graph.reverse)
|
|
if (!files.some((test) => consumers.has(test))) {
|
|
return full('No proven test coverage for changed inputs')
|
|
}
|
|
}
|
|
const affected = unitConsumers([...changed, ...graph.opaque], graph.reverse)
|
|
const selected = files.filter((file) => affected.has(file))
|
|
if (!selected.length) {
|
|
return full('No proven test coverage for changed inputs')
|
|
}
|
|
return {
|
|
files: selected,
|
|
reason: 'Transitive imports plus indirect-input consumers',
|
|
full: false
|
|
}
|
|
}
|
|
|
|
// A concurrency slot is charged per job, not per core, so eight 6.5-minute shards cost eight of
|
|
// the account's slots and made the unit matrix 68% of daily slot demand. Against the checked-in
|
|
// baseline, five shards each carry 24.7 test-minutes over four workers plus ~2.6 minutes of fixed
|
|
// setup, so ~8.8 minutes -- less than the 10.5-minute p95 queue the oversharding was causing.
|
|
export const FULL_SHARD_COUNT = 5
|
|
|
|
export function planUnitSelection({ files, changed, graph, timings, event, mode = 'shadow' }) {
|
|
const candidate = selectUnitFiles(files, changed, graph)
|
|
const selected = mode === 'selected' && event?.pull_request?.draft === true && !candidate.full
|
|
const executionFiles = selected ? candidate.files : files
|
|
const totalMs = balanceFiles(executionFiles, 1, timings).shards[0].durationMs
|
|
const count = selected
|
|
? Math.max(1, Math.min(FULL_SHARD_COUNT, Math.ceil(totalMs / 900_000), executionFiles.length))
|
|
: FULL_SHARD_COUNT
|
|
return {
|
|
version: 1,
|
|
mode: selected ? 'selected' : 'shadow',
|
|
selectionAvailable: !candidate.full,
|
|
reason: candidate.reason,
|
|
files,
|
|
candidateFiles: candidate.files,
|
|
executionFiles,
|
|
shards: Array.from({ length: count }, (_, index) => ({ index: index + 1, count }))
|
|
}
|
|
}
|
|
|
|
export function auditUnitSelection(plan, results) {
|
|
const candidates = new Set(plan.candidateFiles)
|
|
const omittedFailures = Object.entries(results)
|
|
.filter(([file, state]) => state === 'failed' && !candidates.has(file))
|
|
.map(([file]) => file)
|
|
return {
|
|
mode: plan.mode,
|
|
discovered: plan.files.length,
|
|
candidate: candidates.size,
|
|
executed: Object.keys(results).length,
|
|
omittedFailures
|
|
}
|
|
}
|