Spark independent cluster dynamic online and offline worker node

（ One ） Offline Worker node

My operation ： Just turn it off ……
Anyway, the work will naturally be handed over to other nodes .

Of course, it should be like this , Log in to the node host that needs to be offline ：
Execute script to shut down the machine Worker：

$> $SPARK_HOME/sbin/stop-worker.sh

Notice there's another one stop-workers.sh, It is used to close all Worker Of , Don't get me wrong .
See clearly one more s！！！

If you are in the Master I accidentally executed this multiple S Script for （Master To all Worker It must be configured ssh Key login ）,
Then it's all closed ……

Before shutdown ：

After shutting down ：

After a while DEAD Of Worker Will disappear （ The cluster must take some time to see if it can reconnect ）.

.

（ Two ） go online Worker node

Log in to the node host that needs to go online ：
Execute script to start the machine Worker, And connect to Spark Master：

$> $SPARK_HOME/sbin/start-worker.sh spark://vm00:7077

Notice there's another one start-workers.sh Script . The difference is the same as above .

Last , If it's a new one Worker, Don't forget it. $SPARK_CONF_DIR Editor inside workers（ or slaves）.
Add new Worker The host name , It is convenient for all clusters to start together next time .

by the way , A new start Worker There are new ones ID.
Uh …… End .