SparkyR-rsparkling as_h2o_frame()错误java.lang.IllegalArgumentException:不支持的参数:(spark.dynamicAllocation.enabled,true)
我试图在SparkyR-rsparkling as_h2o_frame()错误java.lang.IllegalArgumentException:不支持的参数:(spark.dynamicAllocation.enabled,true),r,hadoop,h2o,sparklyr,R,Hadoop,H2o,Sparklyr,我试图在sparkyr会话期间通过H2o(使用库rsparkling)使用一些机器学习功能。我正在使用hadoop集群 考虑以下示例: library(dplyr) library(sparklyr) library(rsparkling) library(h2o) #configure the spark session and connect sc = spark_connect(master = 'yarn-client', spark_home =
sparkyr
会话期间通过H2o(使用库rsparkling
)使用一些机器学习功能。我正在使用hadoop集群
考虑以下示例:
library(dplyr)
library(sparklyr)
library(rsparkling)
library(h2o)
#configure the spark session and connect
sc = spark_connect(master = 'yarn-client',
spark_home = '/usr/hdp/current/spark-client',
app_name = 'sparklyr',
config = list(
"sparklyr.shell.executor-memory" = "1G",
"sparklyr.shell.driver-memory" = "4G",
"spark.driver.maxResultSize" = "2G" # may need to transfer a lot of data into R
)
)
mtcars_tbl <- copy_to(sc, mtcars, "mtcars")
mtcars_hf <- as_h2o_frame(sc=sc,x=mtcars_tbl,name='h_cars')
库(dplyr)
图书馆(年)
图书馆(公园)
图书馆(h2o)
#配置spark会话并连接
sc=spark_connect(主机=‘纱线客户机’,
spark_home='/usr/hdp/current/spark client',
应用程序名称='Sparkyr',
配置=列表(
“sparklyr.shell.executor内存”=“1G”,
“sparklyr.shell.driver内存”=“4G”,
“spark.driver.maxResultSize”=“2G”#可能需要将大量数据传输到R
)
)
mtcars\u tbl目前,起泡水/RSparkling不支持动态火花束。因此,您只需禁用它:
config=list(“spark.dynamicAllocation.enabled”=“false”)
对于Python用户:
conf = H2OConf(spark).set('spark.dynamicAllocation.enabled', False) # Default of True causes this error: IllegalArgumentException: 'Unsupported argument: (spark.dynamicAllocation.enabled,true)'
hc = H2OContext.getOrCreate(spark, conf)
conf = H2OConf(spark).set('spark.dynamicAllocation.enabled', False) # Default of True causes this error: IllegalArgumentException: 'Unsupported argument: (spark.dynamicAllocation.enabled,true)'
hc = H2OContext.getOrCreate(spark, conf)