I do not understand how it is related, we run CSV bulk loader as Ozzie shell script tasks, from the logs I see that when the script starts all the needed env' variable including the classpath that points to CDH libs is ok. Then from the CSV bulk loading it tries to access some file and fail with the message that file not found
org.apache.phoenix.mapreduce.CsvBulkLoadTool: Import job on table=CDRINFO failed due to exception:java.io.FileNotFoundException: File does not exist: hdfs://dc50-dev-bd01:8020/opt/cloudera/parcels/CDH-5.1.0-1.cdh5.1.0.p0.53/lib/hbase/lib/hbase-server-0.98.1-cdh5.1.0.jar do I need to add all CDH file to share lib for the script to work – it does not make sense From: Artem Ervits [mailto:[email protected]] Sent: Tuesday, October 7, 2014 2:25 PM To: [email protected] Subject: RE: bulk loading using OOZIE Is your oozie sharelib setup? Read about it, it will explain what Oozie expects. On Oct 7, 2014 7:18 AM, "Bulvik, Noam" <[email protected]<mailto:[email protected]>> wrote: I checked the file exists on all cluster machine with full permission (it is part of CDH files ) From: Artem Ervits [mailto:[email protected]<mailto:[email protected]>] Sent: Tuesday, October 7, 2014 1:37 PM To: [email protected]<mailto:[email protected]> Subject: Re: bulk loading using OOZIE from one of your earlier log posts, it says it can't find a jar in the directory, please check the jar is there. On Tue, Oct 7, 2014 at 1:53 AM, Bulvik, Noam <[email protected]<mailto:[email protected]>> wrote: We login to ozzie using mapred user. It looks like the ozzie task runs the script as user nobody , when I executed the command line of the script after loing as nobody it works fine so I do not think we need to explicit permission on some files I also do not think that we need to upload the CDH jars to ozzie (AFAIK) – do you? From: Artem Ervits [mailto:[email protected]<mailto:[email protected]>] Sent: Tuesday, October 7, 2014 12:51 AM To: [email protected]<mailto:[email protected]> Subject: Re: bulk loading using OOZIE Double check that oozie user has permissions to read the required files in hdfs, for example, user running oozie wf is oozie then you should upload the jar and any property files to /user/oozie. On Oct 5, 2014 2:41 AM, "Bulvik, Noam" <[email protected]<mailto:[email protected]>> wrote: Hi, We are trying to do periodic bulk loading using OOZIE as scheduler. We impalement script task that should call the bulk loading command line with all needed parameter. When we run the script directly from Linux console it works fine but when we run the OOZIE task it fail saying that jar is missing or we do not have permission to read it. We are using CDH5.1 and the jar it companies about is one of the CDh jars that we checked and it is exists on all machines of the cluster Any idea ? Regards, Noam Information in this e-mail and its attachments is confidential and privileged under the TEOCO confidentiality terms that can be reviewed here<http://www.teoco.com/email-disclaimer>. -- Artem
