reanimator/bee
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 1 Wiki Activity

feat(audit): add platform thermal cycling stress test

Browse Source 
Runs CPU (stressapptest) + GPU stress simultaneously across multiple
load/idle cycles with varying idle durations (120s/60s/30s) to detect
cooling systems that fail to recover under repeated load.

Presets: smoke (~5 min), acceptance (~25 min), overnight (~100 min).
Outputs metrics.csv + summary.txt with per-cycle throttle and fan
spindown analysis, packed as tar.gz.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Michael Chus
2026-03-29 21:57:33 +03:00
parent 407c1cd1c4
commit 7c2a0135d2
6 changed files with 527 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
audit/internal/webui/pages.go
View File

@@ -615,6 +615,10 @@ func renderBurn() string {
<p style="color:var(--muted);font-size:12px;margin-bottom:8px">Google stressapptest saturates CPU, memory and cache buses simultaneously. Env: <code>BEE_SAT_STRESS_SECONDS</code> (default 300), <code>BEE_SAT_STRESS_MB</code> (default auto).</p>
<button class="btn btn-primary" onclick="runBurnIn('sat-stress')">&#9654; Start SAT Stress</button>
</div></div>
<div class="card"><div class="card-head">Platform Thermal Cycling</div><div class="card-body">
<p style="color:var(--muted);font-size:12px;margin-bottom:8px">Runs CPU + GPU stress simultaneously across multiple load/idle cycles with varying durations. Detects cooling systems that fail to recover under repeated load cycles. Smoke: 2 cycles ~5 min. Acceptance: 4 cycles ~25 min.</p>
<button class="btn btn-primary" onclick="runBurnIn('platform-stress')">&#9654; Start Thermal Cycling</button>
</div></div>
</div>
<div id="bi-output" style="display:none;margin-top:16px" class="card">
<div class="card-head">Output <span id="bi-title"></span></div>

1
audit/internal/webui/server.go
View File

@@ -215,6 +215,7 @@ func NewHandler(opts HandlerOptions) http.Handler {
mux.HandleFunc("POST /api/sat/amd-stress/run", h.handleAPISATRun("amd-stress"))
mux.HandleFunc("POST /api/sat/memory-stress/run", h.handleAPISATRun("memory-stress"))
mux.HandleFunc("POST /api/sat/sat-stress/run", h.handleAPISATRun("sat-stress"))
mux.HandleFunc("POST /api/sat/platform-stress/run", h.handleAPISATRun("platform-stress"))
mux.HandleFunc("GET /api/sat/stream", h.handleAPISATStream)
mux.HandleFunc("POST /api/sat/abort", h.handleAPISATAbort)

35
audit/internal/webui/tasks.go
View File

@@ -12,6 +12,7 @@ import (
"time"
"bee/audit/internal/app"
"bee/audit/internal/platform"
)
// Task statuses.
@@ -34,7 +35,8 @@ var taskNames = map[string]string{
"amd-bandwidth": "AMD GPU MEM Bandwidth",
"amd-stress": "AMD GPU Burn-in",
"memory-stress": "Memory Burn-in",
"sat-stress": "SAT Stress (stressapptest)",
"sat-stress": "SAT Stress (stressapptest)",
"platform-stress": "Platform Thermal Cycling",
"audit": "Audit",
"install": "Install to Disk",
"install-to-ram": "Install to RAM",
@@ -98,6 +100,34 @@ func resolveBurnPreset(profile string) burnPreset {
}
}
func resolvePlatformStressPreset(profile string) platform.PlatformStressOptions {
switch profile {
case "overnight":
return platform.PlatformStressOptions{Cycles: []platform.PlatformStressCycle{
{LoadSec: 600, IdleSec: 120},
{LoadSec: 600, IdleSec: 60},
{LoadSec: 600, IdleSec: 30},
{LoadSec: 600, IdleSec: 120},
{LoadSec: 600, IdleSec: 60},
{LoadSec: 600, IdleSec: 30},
{LoadSec: 600, IdleSec: 120},
{LoadSec: 600, IdleSec: 60},
}}
case "acceptance":
return platform.PlatformStressOptions{Cycles: []platform.PlatformStressCycle{
{LoadSec: 300, IdleSec: 60},
{LoadSec: 300, IdleSec: 30},
{LoadSec: 300, IdleSec: 60},
{LoadSec: 300, IdleSec: 30},
}}
default: // smoke
return platform.PlatformStressOptions{Cycles: []platform.PlatformStressCycle{
{LoadSec: 90, IdleSec: 60},
{LoadSec: 90, IdleSec: 30},
}}
}
}
// taskQueue manages a priority-ordered list of tasks and runs them one at a time.
type taskQueue struct {
mu sync.Mutex
@@ -410,6 +440,9 @@ func (q *taskQueue) runTask(t *Task, j *jobState, ctx context.Context) {
dur = resolveBurnPreset(t.params.BurnProfile).DurationSec
}
archive, err = runSATStressPackCtx(a, ctx, "", dur, j.append)
case "platform-stress":
opts := resolvePlatformStressPreset(t.params.BurnProfile)
archive, err = a.RunPlatformStress(ctx, "", opts, j.append)
case "audit":
result, e := a.RunAuditNow(q.opts.RuntimeMode)
if e != nil {