sinat_19371493的专栏

time vektor -Djava.io.tmpdir=/San1/VWM/user/hzhong/tmp -Xmx5G run --run-dependencies=true -e environment.yaml etl.fa_master_gen.fa_master | tee log0其中time 用来jis

Workflow在cluster跑的时候经常会出现一些错误。一般yo

outputDefinition:    type: "delimited"    header: false    fieldSeparator: "~"    fieldDelimiter: ""    recordSeparator: "\n"    disableParallelism: false相比于 disableParallelism: true

错误一： Out of MemoryERROR >>>ERROR - OutOfMemoryError: unable to create new native threadERROR - DRException: Failed to start dataflow graph etl\.fa_master_gen\.fa_master - createView-phas

- operation: readView          view: performance_visibility.notification.notification.fa_notification          as: fa_notification        - operation: union          as: notification_union

- IOException: No such file or directoryERROR   - RuntimeException: Unable to extract jar:file:/opt/vektor/lib/datarush-hadoop-cdh4-6.3.1-11.jar!/datarush-hadoop-cdh4/com.pervasive.datarush/hadoop-s

Vektor gen folder里面的文件，需要整个目录进行copy

export VEKTOR_HOME=/San1/Vektor-2.1.0-SNAPSHOT/PATH=$PATH:/San1/Vektor-2.1.0-SNAPSHOT/bin

1. 运行了没有反应：可能是server的空间满了。也可能是workplace

做以下尝试1. 清空browser的缓存2. 清空workspace