[英]Shell script through Oozie
我目前正在创建一个工作流,该工作流将自动通过sqoop导入数据。 我要做的是验证通过此过程导入的行(记录)数是否正确; sqoop提供的validate参数在这里不起作用,因为sqoop作业没有导入单个表。
我创建了一个名为“ validation”的动作,该动作将调用并执行一个称为“ validation”的shell脚本。 该验证脚本执行以下步骤:
但是,当我运行它时,出现以下错误,该错误未提供我所需的信息:
2014-09-22 19:03:59,156 INFO ShellActionExecutor:539 - USER[v523043] GROUP[-] TOKEN[-] APP[voipImportToHDFS] JOB[0000359-140905180027053-oozie-oozi-W] ACTION[0000359-140905180027053-oozie-oozi-W@validation] action completed, external ID [null]
2014-09-22 19:03:59,159 WARN ShellActionExecutor:542 - USER[v523043] GROUP[-] TOKEN[-] APP[voipImportToHDFS] JOB[0000359-140905180027053-oozie-oozi-W] ACTION[0000359-140905180027053-oozie-oozi-W@validation] Launcher ERROR, reason: Main class [org.apache.oozie.action.hadoop.ShellMain], exit code [1]
2014-09-22 19:03:59,177 INFO ActionEndXCommand:539 - USER[v523043] GROUP[-] TOKEN[-] APP[voipImportToHDFS] JOB[0000359-140905180027053-oozie-oozi-W] ACTION[0000359-140905180027053-oozie-oozi-W@validation] end executor for wf action 0000359-140905180027053-oozie-oozi-W with wf job 0000359-140905180027053-oozie-oozi-W
2014-09-22 19:03:59,198 INFO ActionEndXCommand:539 - USER[v523043] GROUP[-] TOKEN[-] APP[voipImportToHDFS] JOB[0000359-140905180027053-oozie-oozi-W] ACTION[0000359-140905180027053-oozie-oozi-W@validation] ERROR is considered as FAILED for SLA
我的验证脚本使用以下命令:
Hadoop的FS-猫...
这里有我忽略的兼容性问题吗? 我是否需要进行其他配置?
我的验证脚本(进行中的版本):
for table in ${tables[*]}
do
#Get the number of records from DB Exadata
verifiedCount=$(sqoop eval --connect $3 --query "SELECT COUNT(*) FROM $4.${tables[table]} WHERE INTRVL_DT = To_Date('$5')" | awk '/([0-9]+)/{print $2}')
#echo "Total Number of Records " $verifiedCount
#Count the number of rows imported
totalRows=0
for ((i=0;i<$mapJobs;i++)) {
count[$i]=$(hadoop fs -cat $6$7/${tables[table]}/$8/$9/$10/part-m-0000$i | wc -l)
totalRows=$((totalRows + ${count[$i]}))
#if value has 0 lines, remove the file from edgenode to limit overhead
if [ ${count[$i]} -eq "0" ]
then
hadoop fs -rmr $6$7/${tables[table]}/$8/$9/$10/part-m-0000$i
echo "Removing..."
fi
}
#echo values
if [ "$totalRows" -eq "$verifiedCount" ]
then
echo "evaluation=true"
evaluation=true
else
echo "evaluation=false"
evaluation=false
exit 40
fi
done
蛮力错误调试使我发现以下行错误:
hadoop fs -rmr $6$7/${tables[table]}/$8/$9/$10/part-m-0000$i
特别是,-rmr不能很好地与环境配合使用,从而导致其崩溃。
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.