# Shell/Python节点执行失败，后台报错session is down
本指导以Shell算子为例。
#### 问题现象
Shell节点运行失败了，实际上Shell脚本运行成功了。
![](https://support.huaweicloud.com/dataartsstudio_faq/zh-cn_image_0000001592188529.png "点击放大")
#### 原因分析
1. 获取Shell节点的运行日志。
   ```
   [2021/11/17 02:00:36 GMT+0800] [INFO] No job-level agency is set, Workspace-level agency is dlg_agency, Execute job use agency dlg_agency, job id is 07572F197E4642E5BE549C2B656F157Ctm7cHkHd
   [2021/11/17 02:00:36 GMT+0800] [DEBUG] ===============================================
   [2021/11/17 02:00:36 GMT+0800] [INFO] Get response from agent when try to submit shell running job :
   [2021/11/17 02:00:36 GMT+0800] [INFO]
   {
   "jobResultList":[
   {
   "jobId":"a567f7f5-3c9e-4dfc-a464-bd477ac5b1ea",
   "status":"created",
   "errorCode":0,
   "failCount":0,
   "result":[
   ]
   }
   ],
   "agentId":"614853ee-c1c6-456d-9aa6-fc84ad1281ed"
   }
   [2021/11/17 02:00:36 GMT+0800] [DEBUG] ===============================================
   [2021/11/17 02:05:56 GMT+0800] [DEBUG] ===============================================
   [2021/11/17 02:05:56 GMT+0800] [INFO] Job Run finish , the raw output is :
   [2021/11/17 02:05:56 GMT+0800] [INFO]
   {
   "jobId":"a567f7f5-3c9e-4dfc-a464-bd477ac5b1ea",
   "status":"failed",
   "errorCode":3427,
   "errorMessage":"Shell script job execute failed.",
   "failCount":0,
   "result":[
   {
   "is_success":false,
   "exeTime":300.609
   }
   ]
   }
   [2021/11/17 02:05:56 GMT+0800] [DEBUG] ===============================================
   [2021/11/17 02:05:56 GMT+0800] [DEBUG] ===============================================
   [2021/11/17 02:05:56 GMT+0800] [INFO] The return code is : [-1].
   [2021/11/17 02:05:56 GMT+0800] [DEBUG] ===============================================
   [2021/11/17 02:05:56 GMT+0800] [INFO] Execute shell script job finished.
   [2021/11/17 02:05:56 GMT+0800] [ERROR] Shell exit code is not 0
   [2021/11/17 02:05:56 GMT+0800] [DEBUG] ===============================================
   [2021/11/17 02:05:56 GMT+0800] [ERROR] Shell script job execute failed. Please contact ECS Service.
   [2021/11/17 02:05:56 GMT+0800] [ERROR] Exception message: RuntimeException: Shell script job execute failed. Please contact ECS Service.
   [2021/11/17 02:05:56 GMT+0800] [ERROR] Root Cause message:RuntimeException: Shell script job execute failed. Please contact ECS Service.
   ```
   
2. 确认其ECS的**sshd_config** 参数如下。
   ![](https://support.huaweicloud.com/dataartsstudio_faq/zh-cn_image_0000001591988549.png)
   原因分析：由于ssh session超时断开了，因此Shell节点失败。
   
 
#### 解决办法
1. 编辑ECS的**/etc/ssh/sshd_config** 文件，添加或者更新如下两个值。
   ClientAliveInterval 300
   ClientAliveCountMax 3
   ![](https://support.huaweicloud.com/dataartsstudio_faq/public_sys-resources/note_3.0-zh-cn.png)
   ClientAliveInterval指定了服务器端向客户端请求消息的时间间隔，默认是0，不发送请求。然而ClientAliveInterval 300表示五分钟发送一次，然后客户端响应，这样就保持长连接了。ClientAliveCountMax的默认值3。ClientAliveCountMax表示服务器发出请求后客户端没有响应的次数达到一定值，就自动断开，正常情况下，客户端会正常响应。
   
2. 修改后，重启ECS的sshd，执行如下命令：
   ```
   restart sshd.service
   ```
   ![](https://support.huaweicloud.com/dataartsstudio_faq/zh-cn_image_0000001541389550.png "点击放大")
   
3. 检查sshd是否启动成功（下图为成功）： ![](https://support.huaweicloud.com/dataartsstudio_faq/zh-cn_image_0000001592333837.png "点击放大")