mirror of
https://github.com/JackChen-me/open-multi-agent.git
synced 2026-10-21 10:40:12 -05:00
Jack Chen
9338bb4dec
* chore(observability): complete OBS-5 release readiness * docs(observability): add v2 architecture to core readme
41 lines
1.9 KiB
JavaScript
41 lines
1.9 KiB
JavaScript
import { execFile } from 'node:child_process'
|
|
import { fileURLToPath } from 'node:url'
|
|
import { join } from 'node:path'
|
|
import { promisify } from 'node:util'
|
|
|
|
const exec = promisify(execFile)
|
|
const root = fileURLToPath(new URL('..', import.meta.url))
|
|
const benchmark = join(root, 'packages/core/benchmarks/observability-sinks.mjs')
|
|
const core = join(root, 'packages/core/dist/index.js')
|
|
const otel = join(root, 'packages/otel/dist/index.js')
|
|
const { stdout } = await exec(process.execPath, ['--expose-gc', benchmark, core, otel], {
|
|
cwd: root,
|
|
env: {
|
|
...process.env,
|
|
OMA_BENCH_ITERATIONS: process.env.OMA_BENCH_ITERATIONS ?? '500',
|
|
OMA_BENCH_ROUNDS: process.env.OMA_BENCH_ROUNDS ?? '5',
|
|
},
|
|
maxBuffer: 20 * 1024 * 1024,
|
|
})
|
|
const result = JSON.parse(stdout)
|
|
const failures = []
|
|
|
|
// CI deliberately uses 10x the dedicated RFC microsecond budgets. This catches
|
|
// order-of-magnitude regressions without turning shared-runner jitter into a
|
|
// release blocker. Dedicated benchmark snapshots use the strict budgets.
|
|
if (result.legacyDispatchP95Micros >= 100) failures.push('legacy dispatch p95 >= 100 us')
|
|
if (result.batchEmitP95Micros >= 200) failures.push('batch emit p95 >= 200 us')
|
|
if (result.otelConvertAndProcessP95Micros >= 500) failures.push('OTel conversion p95 >= 500 us')
|
|
for (const [name, value] of Object.entries(result.overheadVsNoSinkPercent)) {
|
|
if (value >= 1_000) failures.push(`${name} whole-run overhead >= 1000% vs no sink`)
|
|
}
|
|
if (failures.length > 0) throw new Error(`Observability benchmark CI gate failed:\n- ${failures.join('\n- ')}`)
|
|
console.log(JSON.stringify({
|
|
gate: 'pass',
|
|
policy: '10x microsecond budgets and <1000% same-host overhead vs no sink',
|
|
legacyDispatchP95Micros: result.legacyDispatchP95Micros,
|
|
batchEmitP95Micros: result.batchEmitP95Micros,
|
|
otelConvertAndProcessP95Micros: result.otelConvertAndProcessP95Micros,
|
|
overheadVsNoSinkPercent: result.overheadVsNoSinkPercent,
|
|
}, null, 2))
|