在提交新问题之前，请确保您已经在社区中搜索过相关问题，并使用了社区中提供的资源/工具后，仍未找到满意的解决方式。

⚠️ 安全信息提醒：请仔细检查提供的文本内容，确保其不包含敏感数据信息，包括但不限于：

• API 令牌或密钥
• 密码或身份验证凭证
• 私有网址或接口地址
• 个人或机密数据
• ...

在分享配置信息或代码示例时，请将敏感信息脱敏处理，或使用 <TOKEN> 等占位符替代原有内容。

环境信息

例如：
- 操作系统
- 昇腾硬件信息
- CANN软件版本
- 安装的对应软件版本

800IA2 64G环境，Qwen2.5-0.5B-Instruct权重，开启beamsearch，n=8192，maxbatchsize配置为8192，输入1000，输出16，并发100，起服务成功后使用jmeter压测，服务偶现OOM

🐛 问题描述

800IA2 64G环境，Qwen2.5-0.5B-Instruct权重，开启beamsearch，n=8192，maxbatchsize配置为8192，输入1000，输出16，并发100，起服务成功后使用jmeter压测，服务偶现OOM

◆【测试步骤】：
1、config文件配置
{
"Version" : "1.0.0",

"ServerConfig" :
{
    "ipAddress" : "80.48.9.147",
    "managementIpAddress" : "127.0.0.2",
    "port" : 1035,
    "managementPort" : 1036,
    "metricsPort" : 1037,
    "allowAllZeroIpListening" : false,
    "maxLinkNum" : 1000,
    "httpsEnabled" : false,
    "fullTextEnabled" : false,
    "tlsCaPath" : "security/ca/",
    "tlsCaFile" : ["ca.pem"],
    "tlsCert" : "security/certs/server.pem",
    "tlsPk" : "security/keys/server.key.pem",
    "tlsCrlPath" : "security/certs/",
    "tlsCrlFiles" : ["server_crl.pem"],
    "managementTlsCaFile" : ["management_ca.pem"],
    "managementTlsCert" : "security/certs/management/server.pem",
    "managementTlsPk" : "security/keys/management/server.key.pem",
    "managementTlsCrlPath" : "security/management/certs/",
    "managementTlsCrlFiles" : ["server_crl.pem"],
    "metricsTlsCaFile" : ["metrics_ca.pem"],
    "metricsTlsCert" : "security/certs/metrics/server.pem",
    "metricsTlsPk" : "security/keys/metrics/server.key.pem",
    "metricsTlsCrlPath" : "security/metrics/certs/",
    "metricsTlsCrlFiles" : ["server_crl.pem"],
    "inferMode" : "standard",
    "interCommTLSEnabled" : true,
    "interCommPort" : 1121,
    "interCommTlsCaPath" : "security/grpc/ca/",
    "interCommTlsCaFiles" : ["ca.pem"],
    "interCommTlsCert" : "security/grpc/certs/server.pem",
    "interCommPk" : "security/grpc/keys/server.key.pem",
    "interCommTlsCrlPath" : "security/grpc/certs/",
    "interCommTlsCrlFiles" : ["server_crl.pem"],
    "openAiSupport" : "vllm",
    "tokenTimeout" : 3600,
    "e2eTimeout" : 65535,
    "distDPServerEnabled":false,
    "layerwiseDisaggregated" : false,
    "layerwiseDisaggregatedRoleType" : "",
    "layerwiseDisaggregatedMasterIpAddress" : "127.0.0.3",
    "layerwiseDisaggregatedSlaveIpAddress" : ["127.0.0.4"],
    "layerwiseDisaggregatedDataPort" : 10024,
    "layerwiseDisaggregatedCrtlPort" : [10001,10002],
    "HealthCheckConfig" :
    {
        "npuUsageThreshold" : 0
    }
},

"BackendConfig" : {
    "backendName" : "mindieservice_llm_engine",
    "modelInstanceNumber" : 1,
    "npuDeviceIds" : [[0]],
    "tokenizerProcessNumber" : 8,
    "multiNodesInferEnabled" : false,
    "multiNodesInferPort" : 1120,
    "interNodeTLSEnabled" : true,
    "interNodeTlsCaPath" : "security/grpc/ca/",
    "interNodeTlsCaFiles" : ["ca.pem"],
    "interNodeTlsCert" : "security/grpc/certs/server.pem",
    "interNodeTlsPk" : "security/grpc/keys/server.key.pem",
    "interNodeTlsCrlPath" : "security/grpc/certs/",
    "interNodeTlsCrlFiles" : ["server_crl.pem"],
    "kvPoolConfig" : {"backend":"", "configPath":""},
    "ModelDeployConfig" :
    {
        "maxSeqLen" : 2560,
        "maxInputTokenLen" : 2048,
        "truncation" : false,
        "ModelConfig" : [
            {
                "modelInstanceType" : "Standard",
                "modelName" : "qwen",
                "modelWeightPath" : "/data/weights/Qwen2.5-0.5B-Instruct",
                "worldSize" : 1,
                "cpuMemSize" : 0,
                "npuMemSize" : -1,
                "plugin_params": "{\"plugin_type\":\"splitfuse,prefix_cache\"}",
                "backendType" : "atb",
                "trustRemoteCode" : false,
                "async_scheduler_wait_time": 120,
                "kv_trans_timeout": 10,
                "kv_link_timeout": 1080,
                "models" : {
                    "layerwiseDisaggregatedMasterDeviceNum" : 2,
                    "layerwiseDisaggregatedSlaveDeviceNum" : 8
                }
            }
        ]
    },

    "ScheduleConfig" :
    {
        "templateType" : "Mix",
        "templateName" : "Standard_LLM",
        "cacheBlockSize" : 128,

        "maxPrefillBatchSize" : 50,
        "maxPrefillTokens" : 8192,
        "prefillTimeMsPerReq" : 150,
        "prefillPolicyType" : 0,

        "decodeTimeMsPerReq" : 50,
        "decodePolicyType" : 0,

        "maxBatchSize" : 8192,
        "maxIterTimes" : 512,
        "maxPreemptCount" : 0,
        "supportSelectBatch" : false,
        "maxQueueDelayMicroseconds" : 5000,
        "maxFirstTokenWaitTime": 2500
    }
},

"LogConfig": {
    "dynamicLogLevel" : "",
    "dynamicLogLevelValidHours" : 2,
    "dynamicLogLevelValidTime" : ""
},

"EnableDynamicAdjustTimeoutConfig": false

}

欢迎加入社区，感谢您对社区的贡献 🎉!