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

Refactor validate modes, fix benchmark report and IPMI power

Browse Source 
- Replace diag level 1-4 dropdown with Validate/Stress radio buttons
- Validate: dcgmi L2, 60s CPU, 256MB/1p memtester, SMART short
- Stress: dcgmi L3 + targeted_stress in Run All, 30min CPU, 1GB/3p memtester, SMART long/NVMe extended
- Parallel GPU mode: spawn single task for all GPUs instead of splitting per model
- Benchmark table: per-GPU columns for sequential runs, server-wide column for parallel
- Benchmark report converted to Markdown with server model, GPU model, version in header; only steady-state charts
- Fix IPMI power parsing in benchmark (was looking for 'Current Power', correct field is 'Instantaneous power reading')

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Michael Chus
2026-04-08 00:42:12 +03:00
parent dd26e03b2d
commit b2f8626fee
12 changed files with 332 additions and 164 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
audit/internal/webui/tasks.go
View File

@@ -115,10 +115,11 @@ type Task struct {
// taskParams holds optional parameters parsed from the run request.
type taskParams struct {
Duration int `json:"duration,omitempty"`
DiagLevel int `json:"diag_level,omitempty"`
StressMode bool `json:"stress_mode,omitempty"`
GPUIndices []int `json:"gpu_indices,omitempty"`
ExcludeGPUIndices []int `json:"exclude_gpu_indices,omitempty"`
SizeMB int `json:"size_mb,omitempty"`
Passes int `json:"passes,omitempty"`
Loader string `json:"loader,omitempty"`
BurnProfile string `json:"burn_profile,omitempty"`
BenchmarkProfile string `json:"benchmark_profile,omitempty"`
@@ -215,11 +216,11 @@ var globalQueue = &taskQueue{trigger: make(chan struct{}, 1)}
const maxTaskHistory = 50
var (
runMemoryAcceptancePackCtx = func(a *app.App, ctx context.Context, baseDir string, logFunc func(string)) (string, error) {
return a.RunMemoryAcceptancePackCtx(ctx, baseDir, logFunc)
runMemoryAcceptancePackCtx = func(a *app.App, ctx context.Context, baseDir string, sizeMB, passes int, logFunc func(string)) (string, error) {
return a.RunMemoryAcceptancePackCtx(ctx, baseDir, sizeMB, passes, logFunc)
}
runStorageAcceptancePackCtx = func(a *app.App, ctx context.Context, baseDir string, logFunc func(string)) (string, error) {
return a.RunStorageAcceptancePackCtx(ctx, baseDir, logFunc)
runStorageAcceptancePackCtx = func(a *app.App, ctx context.Context, baseDir string, extended bool, logFunc func(string)) (string, error) {
return a.RunStorageAcceptancePackCtx(ctx, baseDir, extended, logFunc)
}
runCPUAcceptancePackCtx = func(a *app.App, ctx context.Context, baseDir string, durationSec int, logFunc func(string)) (string, error) {
return a.RunCPUAcceptancePackCtx(ctx, baseDir, durationSec, logFunc)
@@ -552,7 +553,10 @@ func (q *taskQueue) runTask(t *Task, j *jobState, ctx context.Context) {
err = fmt.Errorf("app not configured")
break
}
diagLevel := t.params.DiagLevel
diagLevel := 2
if t.params.StressMode {
diagLevel = 3
}
if len(t.params.GPUIndices) > 0 || diagLevel > 0 {
result, e := a.RunNvidiaAcceptancePackWithOptions(
ctx, "", diagLevel, t.params.GPUIndices, j.append,
@@ -658,13 +662,17 @@ func (q *taskQueue) runTask(t *Task, j *jobState, ctx context.Context) {
err = fmt.Errorf("app not configured")
break
}
archive, err = runMemoryAcceptancePackCtx(a, ctx, "", j.append)
sizeMB, passes := 256, 1
if t.params.StressMode {
sizeMB, passes = 1024, 3
}
archive, err = runMemoryAcceptancePackCtx(a, ctx, "", sizeMB, passes, j.append)
case "storage":
if a == nil {
err = fmt.Errorf("app not configured")
break
}
archive, err = runStorageAcceptancePackCtx(a, ctx, "", j.append)
archive, err = runStorageAcceptancePackCtx(a, ctx, "", t.params.StressMode, j.append)
case "cpu":
if a == nil {
err = fmt.Errorf("app not configured")
@@ -675,7 +683,11 @@ func (q *taskQueue) runTask(t *Task, j *jobState, ctx context.Context) {
dur = resolveBurnPreset(t.params.BurnProfile).DurationSec
}
if dur <= 0 {
dur = 60
if t.params.StressMode {
dur = 1800
} else {
dur = 60
}
}
j.append(fmt.Sprintf("CPU stress duration: %ds", dur))
archive, err = runCPUAcceptancePackCtx(a, ctx, "", dur, j.append)