<spark> error: View process after initiating spark, master and worker process conflict in process

After starting Hadoop and then starting Spark JPS, the master process and worker process are found to be present, and a half-day configuration file is debugged.

The test found that when I shut down Hadoop the worker process still exists,

However, when I shut down spark again and then JPS, I found that the worker process still exists.

Then remembered in the ~/spark/conf/slaves configuration of slave1 slave2 above also has a localhost, directly delete localhost, and then Kill-s 9 worker process.

The first Test solved the error, but do not know whether it is temporary, if the problem is still unresolved or there is a better solution, please tell me.

<spark> error: View process after initiating spark, master and worker process conflict in process