Solve the problem that Flink cannot be closed normally after startup

Use command stop-cluster.sh close flink The following error occurs when clustering ,

No taskexecutor daemon to stop on host tanghc-X550JX.
No standalonesession daemon to stop on host tanghc-X550JX.

jps Look at the process and find ,flink It's on ,8081 Port viewing is also normal ,

Why can't it shut down normally , Not yet prompted taskexecutor and standalonesession What about the process ？

I think of the past spark Similar problems have occurred in clusters , as a result of spark Starting will start the process ID Save to a file , If the configuration is not modified, the file is saved to the system by default /tmp Under the table of contents , Because it is a temporary directory , Will be cleaned up by the system , So the storage ID I can't find , There is no way to shut down the cluster .

Go to the /tmp View files in directory , I found it flink Of PID file

So it's easy to do , As long as this configuration item is changed, it will be all alive .
according to spark First of all, I enter FLINK_HOME Of conf Under the table of contents

open flink-conf.yaml, Find out , Can't find PID Configuration options , I didn't admit defeat , I read it line by line again , Not really. , Don't flink Do not modify ？？ I thought with my thigh , Back to my question , impossible ！ It must be modifiable , It suddenly occurred to me , Or look at the startup script ？

Hey, hey, hey ～ No, ！ But English is super strong （la） harm （ji） I found this

stay FLINK_HOME/bin/ I saw it under the directory config.sh file
vim open

I saw the familiar configuration items ：××××××_PID_DIR
Sure enough, the designated is /tmp, Modify this configuration item to the appropriate location （ Bloggers directly in FLINK_HOME We built a tmp Catalog ）

However, the process that cannot be closed before can only be forced Kill It fell off .
Successfully shut down after restart