mirror of
https://gitee.com/fit2cloud-feizhiyun/1Panel.git
synced 2026-11-20 07:49:30 +08:00
feat: add ai benchmark
This commit is contained in:
@@ -471,6 +471,7 @@ TaskScan: "Scan"
|
||||
# task - ai
|
||||
OllamaModelPull: 'Pull Ollama model {{ .name }}'
|
||||
OllamaModelSize: 'Get Ollama model size: {{ .name }}'
|
||||
AIBenchmarkRun: 'Run AI benchmark'
|
||||
|
||||
# task-snapshot
|
||||
Snapshot: 'Snapshot'
|
||||
|
||||
@@ -471,6 +471,7 @@ TaskScan: "扫描"
|
||||
# task - ai
|
||||
OllamaModelPull: "拉取 Ollama 模型 {{ .name }}"
|
||||
OllamaModelSize: "获取 Ollama 模型 {{ .name }} 大小"
|
||||
AIBenchmarkRun: "执行 AI 基准测试"
|
||||
|
||||
# task - snapshot
|
||||
Snapshot: "快照"
|
||||
|
||||
@@ -57,6 +57,7 @@ const (
|
||||
TaskCreateCluster = "TaskCreateCluster"
|
||||
TaskBackup = "TaskBackup"
|
||||
TaskPush = "TaskPush"
|
||||
TaskExec = "TaskExec"
|
||||
)
|
||||
|
||||
const (
|
||||
@@ -66,6 +67,7 @@ const (
|
||||
TaskScopeAppBackup = "AppBackup"
|
||||
TaskScopeCluster = "Cluster"
|
||||
TaskScopeAppInstall = "AppInstallTask"
|
||||
TaskScopeAI = "AI"
|
||||
)
|
||||
|
||||
func GetTaskName(resourceName, operate, scope string) string {
|
||||
|
||||
@@ -1,8 +1,9 @@
|
||||
package constant
|
||||
|
||||
const (
|
||||
StatusSuccess = "Success"
|
||||
StatusFailed = "Failed"
|
||||
StatusSuccess = "Success"
|
||||
StatusFailed = "Failed"
|
||||
StatusCanceled = "Canceled"
|
||||
|
||||
// node
|
||||
StatusWaiting = "Waiting"
|
||||
|
||||
@@ -131,6 +131,7 @@ TaskUpgrade: "Upgrade"
|
||||
TaskSync: 'Sync'
|
||||
TaskSyncForNode: "Sync node data"
|
||||
TaskBackup: "Backup"
|
||||
AIBenchmarkRun: "Run AI benchmark"
|
||||
SuccessStatus: "{{ .name }} succeeded"
|
||||
FailedStatus: "{{ .name }} failed {{ .err }}"
|
||||
Start: "Start"
|
||||
|
||||
@@ -130,6 +130,7 @@ TaskUpgrade: "升级"
|
||||
TaskSync: "同步"
|
||||
TaskSyncForNode: "同步节点数据"
|
||||
TaskBackup: "备份"
|
||||
AIBenchmarkRun: "执行 AI 基准测试"
|
||||
SuccessStatus: "{{ .name }} 成功"
|
||||
FailedStatus: "{{ .name }} 失败 {{ .err }}"
|
||||
Start: "开始"
|
||||
|
||||
@@ -86,6 +86,18 @@ func LoadMenus() string {
|
||||
}
|
||||
if global.CONF.Base.IsEnterprise {
|
||||
for i := range item {
|
||||
if item[i].Label == "AI-Menu" {
|
||||
item[i].Children = UpsertMenuByLabel(item[i].Children, dto.ShowMenu{
|
||||
ID: "45",
|
||||
Disabled: false,
|
||||
Title: "aiTools.benchmark.title",
|
||||
IsShow: true,
|
||||
Label: "AIBenchmark",
|
||||
Path: "/ai/benchmark",
|
||||
Sort: 150,
|
||||
}, "AIModel")
|
||||
continue
|
||||
}
|
||||
if item[i].Label != "Xpack-Menu" {
|
||||
continue
|
||||
}
|
||||
@@ -112,6 +124,7 @@ func MenuSort() []dto.MenuLabelSort {
|
||||
{Label: "AI-Menu", Sort: 300},
|
||||
{Label: "Agents", Sort: 50},
|
||||
{Label: "AIModel", Sort: 100},
|
||||
{Label: "AIBenchmark", Sort: 150},
|
||||
{Label: "MCPServer", Sort: 200},
|
||||
{Label: "GPU", Sort: 300},
|
||||
{Label: "Website-Menu", Sort: 400},
|
||||
|
||||
@@ -39,6 +39,7 @@ func Init() {
|
||||
migrations.AddDocSourceSetting,
|
||||
migrations.AddAppStoreInstallAllowPortSetting,
|
||||
migrations.AddUserManagementMenu,
|
||||
migrations.AddAIBenchmarkMenu,
|
||||
migrations.AddOperationLogUser,
|
||||
migrations.AddIsOfflineSetting,
|
||||
})
|
||||
|
||||
@@ -953,6 +953,15 @@ func buildAiMenuChildren(children []dto.ShowMenu) []dto.ShowMenu {
|
||||
Path: "/ai/model/account",
|
||||
Sort: 100,
|
||||
}, "AIModel", "OllamaModel"),
|
||||
normalizeAiMenuChild(children, dto.ShowMenu{
|
||||
ID: "45",
|
||||
Label: "AIBenchmark",
|
||||
Disabled: false,
|
||||
IsShow: true,
|
||||
Title: "aiTools.benchmark.title",
|
||||
Path: "/ai/benchmark",
|
||||
Sort: 150,
|
||||
}, "AIBenchmark"),
|
||||
normalizeAiMenuChild(children, dto.ShowMenu{
|
||||
ID: "42",
|
||||
Label: "MCPServer",
|
||||
@@ -993,6 +1002,55 @@ func normalizeAiMenuChild(children []dto.ShowMenu, fallback dto.ShowMenu, labels
|
||||
return fallback
|
||||
}
|
||||
|
||||
var AddAIBenchmarkMenu = &gormigrate.Migration{
|
||||
ID: "20260429-add-ai-benchmark-menu",
|
||||
Migrate: func(tx *gorm.DB) error {
|
||||
if !global.CONF.Base.IsEnterprise {
|
||||
return nil
|
||||
}
|
||||
var menuJSON string
|
||||
if err := tx.Model(&model.Setting{}).Where("key = ?", "HideMenu").Pluck("value", &menuJSON).Error; err != nil {
|
||||
return err
|
||||
}
|
||||
if menuJSON == "" {
|
||||
menuJSON = helper.LoadMenus()
|
||||
}
|
||||
|
||||
var menus []dto.ShowMenu
|
||||
if err := json.Unmarshal([]byte(menuJSON), &menus); err != nil {
|
||||
return tx.Model(&model.Setting{}).
|
||||
Where("key = ?", "HideMenu").
|
||||
Update("value", helper.LoadMenus()).Error
|
||||
}
|
||||
|
||||
newItem := dto.ShowMenu{
|
||||
ID: "45",
|
||||
Disabled: false,
|
||||
Title: "aiTools.benchmark.title",
|
||||
IsShow: true,
|
||||
Label: "AIBenchmark",
|
||||
Path: "/ai/benchmark",
|
||||
Sort: 150,
|
||||
}
|
||||
|
||||
for i := range menus {
|
||||
if menus[i].Label != "AI-Menu" {
|
||||
continue
|
||||
}
|
||||
menus[i].Children = helper.UpsertMenuByLabel(menus[i].Children, newItem, "AIModel")
|
||||
break
|
||||
}
|
||||
|
||||
updatedJSON, err := json.Marshal(menus)
|
||||
if err != nil {
|
||||
return tx.Model(&model.Setting{}).
|
||||
Where("key = ?", "HideMenu").
|
||||
Update("value", helper.LoadMenus()).Error
|
||||
}
|
||||
return tx.Model(&model.Setting{}).Where("key = ?", "HideMenu").Update("value", string(updatedJSON)).Error
|
||||
},
|
||||
}
|
||||
|
||||
var AddUserManagementMenu = &gormigrate.Migration{
|
||||
ID: "20260415-add-user-management-menu",
|
||||
Migrate: func(tx *gorm.DB) error {
|
||||
|
||||
+18
-2
@@ -22,6 +22,7 @@ import (
|
||||
const maxStreamOutputCapture = 64 * 1024
|
||||
|
||||
type CommandHelper struct {
|
||||
context context.Context
|
||||
workDir string
|
||||
outputFile string
|
||||
env []string
|
||||
@@ -167,7 +168,10 @@ func (c *CommandHelper) RunPipe(commands ...PipeCommand) (string, error) {
|
||||
}
|
||||
|
||||
func (c *CommandHelper) pipeContext() (context.Context, context.CancelFunc) {
|
||||
ctx := context.Background()
|
||||
ctx := c.context
|
||||
if ctx == nil {
|
||||
ctx = context.Background()
|
||||
}
|
||||
if c.timeout == 0 {
|
||||
return ctx, nil
|
||||
}
|
||||
@@ -243,9 +247,16 @@ func (c *CommandHelper) run(name string, arg ...string) (string, error) {
|
||||
var cancel context.CancelFunc
|
||||
|
||||
if c.timeout != 0 {
|
||||
ctx, cancel = context.WithTimeout(context.Background(), c.timeout)
|
||||
if c.context == nil {
|
||||
ctx, cancel = context.WithTimeout(context.Background(), c.timeout)
|
||||
} else {
|
||||
ctx, cancel = context.WithTimeout(c.context, c.timeout)
|
||||
}
|
||||
defer cancel()
|
||||
cmd = exec.CommandContext(ctx, name, arg...)
|
||||
} else if c.context != nil {
|
||||
ctx = c.context
|
||||
cmd = exec.CommandContext(ctx, name, arg...)
|
||||
} else {
|
||||
cmd = exec.Command(name, arg...)
|
||||
}
|
||||
@@ -337,6 +348,11 @@ func WithOutputFile(outputFile string) Option {
|
||||
s.outputFile = outputFile
|
||||
}
|
||||
}
|
||||
func WithContext(ctx context.Context) Option {
|
||||
return func(s *CommandHelper) {
|
||||
s.context = ctx
|
||||
}
|
||||
}
|
||||
func WithTimeout(timeout time.Duration) Option {
|
||||
return func(s *CommandHelper) {
|
||||
s.timeout = timeout
|
||||
|
||||
@@ -914,12 +914,15 @@ const message = {
|
||||
failureCount: 'Failures',
|
||||
lastError: 'Last Error',
|
||||
modelMap: 'Model Mapping',
|
||||
apiKeys: 'API Keys',
|
||||
apiKey: 'API Key',
|
||||
apiKeyKeepEmpty: 'API Key (leave empty to keep unchanged)',
|
||||
apiKeyOnceHint: 'The API Key is shown only once after creation. Copy and save it now.',
|
||||
apiKeyOnceSave: 'The API Key is shown only once. Save it now:',
|
||||
apiKeyReset: 'Reset API Key (leave empty to keep unchanged)',
|
||||
qpsLimit: 'QPS Limit',
|
||||
unlimited: 'Unlimited',
|
||||
userGroup: 'User Group',
|
||||
userGroups: 'User Groups',
|
||||
defaultGroup: 'Default',
|
||||
apiKeyCount: 'API Keys',
|
||||
lastUsedAt: 'Last Used',
|
||||
settingTitle: 'AI Proxy Settings',
|
||||
lbPolicy: 'Load Balancing Policy',
|
||||
@@ -947,7 +950,8 @@ const message = {
|
||||
serviceOperateConfirm: 'Confirm to {0} the AI Proxy service?',
|
||||
deleteBackendConfirm: 'Delete model account {0}?',
|
||||
deleteBackendTitle: 'Delete Model Account',
|
||||
deleteUserConfirm: 'Delete user {0}?',
|
||||
deleteUserConfirm: 'Delete API Key {0}?',
|
||||
deleteGroupConfirm: 'Delete user group {0}?',
|
||||
userFallback: 'User {0}',
|
||||
requestModel: 'Request Model',
|
||||
upstreamModel: 'Upstream Model',
|
||||
@@ -983,6 +987,44 @@ const message = {
|
||||
validationModelMapEmpty: 'Model mapping names cannot be empty',
|
||||
validationModelMapDuplicate: 'Request model {0} is duplicated',
|
||||
},
|
||||
benchmark: {
|
||||
title: 'Benchmark',
|
||||
create: 'Create',
|
||||
launchCommand: 'Launch Command',
|
||||
retest: 'Retest',
|
||||
statusRunning: 'Running',
|
||||
statusWaiting: 'Waiting',
|
||||
base: 'Basic Information',
|
||||
backend: 'Backend',
|
||||
baseUrl: 'Base URL',
|
||||
endpoint: 'Endpoint',
|
||||
tokenizer: 'Tokenizer',
|
||||
tokenizerPlaceholder:
|
||||
'Select or enter a local tokenizer directory, e.g. /opt/1panel/tokenizers/DeepSeek-V3',
|
||||
runConfig: 'Run Config',
|
||||
rawResult: 'Raw Result',
|
||||
contextTokens: 'Context Length',
|
||||
contextTokensHelper: 'Input token limit plus output token limit for this benchmark.',
|
||||
outputThroughput: 'Output Throughput',
|
||||
outputThroughputHelper: 'Generated output tokens per second. Higher means faster generation.',
|
||||
totalThroughput: 'Total Throughput',
|
||||
totalThroughputHelper: 'Input and output tokens processed per second. Used to measure overall capacity.',
|
||||
firstTokenLatency: 'First Token Latency',
|
||||
firstTokenLatencyHelper:
|
||||
'Time from sending the request to receiving the first token. Lower means faster response.',
|
||||
inputTokens: 'Input Tokens',
|
||||
outputTokens: 'Output Tokens',
|
||||
numPrompts: 'Prompts',
|
||||
concurrency: 'Concurrency',
|
||||
requestRate: 'Request Rate',
|
||||
requestRateUnlimited: 'Unlimited (max throughput)',
|
||||
requestRateCustom: 'Custom QPS',
|
||||
requestRateCustomPlaceholder: 'Requests per second, e.g. 2.5',
|
||||
timeout: 'Timeout (seconds)',
|
||||
image: 'vLLM Image',
|
||||
ignoreEos: 'Ignore EOS',
|
||||
extraHeaders: 'Extra Headers',
|
||||
},
|
||||
gpu: {
|
||||
gpu: 'GPU Monitoring',
|
||||
gpuHelper: 'The system did not detect NVIDIA-SMI or XPU-SMI commands. Please check and try again!',
|
||||
|
||||
@@ -858,12 +858,15 @@ const message = {
|
||||
failureCount: '失败次数',
|
||||
lastError: '最近错误',
|
||||
modelMap: '模型映射',
|
||||
apiKeys: 'API Key',
|
||||
apiKey: 'API Key',
|
||||
apiKeyKeepEmpty: 'API Key(留空不改)',
|
||||
apiKeyOnceHint: 'API Key 仅创建时显示一次,请及时复制保存。',
|
||||
apiKeyOnceSave: 'API Key 只显示一次,请立即保存:',
|
||||
apiKeyReset: '重置 API Key(留空不改)',
|
||||
qpsLimit: 'QPS 限制',
|
||||
unlimited: '不限',
|
||||
userGroup: '用户组',
|
||||
userGroups: '用户组',
|
||||
defaultGroup: '默认',
|
||||
apiKeyCount: 'API Key 数',
|
||||
lastUsedAt: '最近使用时间',
|
||||
settingTitle: 'AI 代理设置',
|
||||
lbPolicy: '负载策略',
|
||||
@@ -889,7 +892,8 @@ const message = {
|
||||
serviceOperateConfirm: '确认{0} AI 代理服务?',
|
||||
deleteBackendConfirm: '确认删除模型账号 {0}?',
|
||||
deleteBackendTitle: '删除模型账号',
|
||||
deleteUserConfirm: '确认删除用户 {0}?',
|
||||
deleteUserConfirm: '确认删除 API Key {0}?',
|
||||
deleteGroupConfirm: '确认删除用户组 {0}?',
|
||||
userFallback: '用户 {0}',
|
||||
requestModel: '请求模型',
|
||||
upstreamModel: '上游模型',
|
||||
@@ -925,6 +929,42 @@ const message = {
|
||||
validationModelMapEmpty: '模型映射的模型名不能为空',
|
||||
validationModelMapDuplicate: '请求模型 {0} 重复',
|
||||
},
|
||||
benchmark: {
|
||||
title: '基准测试',
|
||||
create: '创建',
|
||||
launchCommand: '启动命令',
|
||||
retest: '重新测试',
|
||||
statusRunning: '运行中',
|
||||
statusWaiting: '等待中',
|
||||
base: '基础信息',
|
||||
backend: '后端',
|
||||
baseUrl: '服务地址',
|
||||
endpoint: '接口路径',
|
||||
tokenizer: '分词器',
|
||||
tokenizerPlaceholder: '请选择或输入本地分词器目录,例如 /opt/1panel/tokenizers/DeepSeek-V3',
|
||||
runConfig: '运行配置',
|
||||
rawResult: '原始结果',
|
||||
contextTokens: '上下文长度',
|
||||
contextTokensHelper: '本次测试的输入 Token 与输出 Token 上限之和。',
|
||||
outputThroughput: '输出吞吐',
|
||||
outputThroughputHelper: '模型每秒生成的输出 Token 数,越高表示生成速度越快。',
|
||||
totalThroughput: '总吞吐',
|
||||
totalThroughputHelper: '每秒处理的输入与输出 Token 总数,用于衡量整体处理能力。',
|
||||
firstTokenLatency: '首 Token 延迟',
|
||||
firstTokenLatencyHelper: '从请求发出到收到第一个 Token 的时间,越低表示响应越快。',
|
||||
inputTokens: '输入 Token',
|
||||
outputTokens: '输出 Token',
|
||||
numPrompts: '请求数量',
|
||||
concurrency: '并发数',
|
||||
requestRate: '请求速率',
|
||||
requestRateUnlimited: '不限速(最大吞吐)',
|
||||
requestRateCustom: '自定义 QPS',
|
||||
requestRateCustomPlaceholder: '请输入每秒请求数,如 2.5',
|
||||
timeout: '超时时间(秒)',
|
||||
image: 'vLLM 镜像',
|
||||
ignoreEos: '忽略 EOS',
|
||||
extraHeaders: '额外请求头',
|
||||
},
|
||||
gpu: {
|
||||
gpu: 'GPU 监控',
|
||||
gpuHelper: '当前系统未检测到 NVIDIA-SMI 或 XPU-SMI 指令,请检查后重试!',
|
||||
|
||||
Reference in New Issue
Block a user