解决mlflow 无法打开的问题

解决mlflow 无法打开的问题 项目场景：ubuntu上使用mlflow追踪模型训练参数时，无法打开mlflow的场景

提示：之前使用mlflow ui可以追踪模型参数，但是在关掉对应窗口后就出现了下面的情形，关闭方式采用“Ctrl+c”或直接关闭浏览器所致

---

报错如下：

[2021-12-27 17:11:42 +0800] [19609] [INFO] Starting gunicorn 20.1.0
[2021-12-27 17:11:42 +0800] [19609] [ERROR] Connection in use: ('127.0.0.1', 5000)
[2021-12-27 17:11:42 +0800] [19609] [ERROR] Retrying in 1 second.
[2021-12-27 17:11:43 +0800] [19609] [ERROR] Connection in use: ('127.0.0.1', 5000)
[2021-12-27 17:11:43 +0800] [19609] [ERROR] Retrying in 1 second.
[2021-12-27 17:11:44 +0800] [19609] [ERROR] Connection in use: ('127.0.0.1', 5000)
[2021-12-27 17:11:44 +0800] [19609] [ERROR] Retrying in 1 second.
[2021-12-27 17:11:45 +0800] [19609] [ERROR] Connection in use: ('127.0.0.1', 5000)
[2021-12-27 17:11:45 +0800] [19609] [ERROR] Retrying in 1 second.
[2021-12-27 17:11:46 +0800] [19609] [ERROR] Connection in use: ('127.0.0.1', 5000)
[2021-12-27 17:11:46 +0800] [19609] [ERROR] Retrying in 1 second.
[2021-12-27 17:11:47 +0800] [19609] [ERROR] Can't connect to ('127.0.0.1', 5000)
Running the mlflow server failed. Please see the logs above for details.

原因分析：

提示：mlflow对应的进程仍然被占用着，需要手动kill

---

解决方案：

使用ps -A | grep gunicorn查看gunicorn的进程，然后使用 kill [PID]手动kill掉gunicorn就可以了，在重新启用mlflow ui就可以了
mHandler.obtainMessage(READ_DATA, bytes, -1, buffer).sendToTarget();换成 mHandler.sendMessage()。

参考：How to safely shutdown mlflow ui