Flink run 方式提交（生产环境在yarn集群上直接提交的方式）

1、案例

#!/bin/bash
########################
. /etc/profile
. ~/.bash_profile
########################


SCRIPT_NAME=$(readlink -f "$0")
hostname=`hostname`
dir=`dirname ${SCRIPT_NAME}`
cd $dir

nohup \
flink run \
-d \
-p 4 \
-m yarn-cluster \
-yjm 1G \
-ytm 4G \
-yn 3 \
-ys 2 \
-ynm "LM_VV_STATISTICS" \
-yqu root.yarn \
-c com.stream.lm.vv.LmVvStatistics \
$dir/flink-stream-1.0-SNAPSHOT.jar \
>> $dir/logs/application.log 2>&1 &

2、参数的详细含义

 flink run -m yarn-cluster对于yarn的资源的命令参数如下：

# -m/--jobmanager : yarn-cluster集群
# -yd/--yarndetached : 后台
# -yjm/--yarnjobManager : jobmanager的内存
# -ytm/--yarntaskManager : taskmanager的内存
# -yid/--yarnapplicationId : job依附的applicationId
# -ynm/--yarnname : application的名称
# -ys/--yarnslots : 分配的slots个数

flink run -m yarn-cluster -yd -yjm 1024m -ytm 1024m -ynm  -ys 1