Page 1 of 1
To capture job statistics
Posted: Thu Aug 16, 2007 2:01 pm
by parsi_cnu
how to read the job log and to capture the statistics
like datastage job name,start time,end time ,project, duration,sucess etc
Posted: Thu Aug 16, 2007 2:25 pm
by kduke
Do search for EtlStats. It is a bunch of jobs I give away on my tips page.
Re: To capture job statistics
Posted: Fri Aug 17, 2007 12:14 am
by ghila
parsi_cnu wrote:how to read the job log and to capture the statistics
like datastage job name,start time,end time ,project, duration,sucess etc
You can also use the "dsjob -report" command ( check the Advanced Developer's Guide for a list of all available options ).
I do prefer to use the XML option because it can easily be formated in HTML and that option lets you have the most detailed informations ( like PID of processes, calculated CPU time... ).
Posted: Fri Aug 17, 2007 7:06 am
by kduke
The main job in EtlStats uses dsjob -report XML to load these stats into tables. The DDL for ETL_JOB_HIST, ETL_ROW_HIST and ETL_PARAM_HIST is included in the zip file. The bugs in XML forced me to process the XML file 3 times. This main job is called DSJobReportDb.
If you call job DSJobReportDbDriver at the end of your sequences then it will get your row counts for all jobs in this sequence. It will call EtlChkSeqJobsNotRun and email you the log in an html attachment of all jobs which failed.
You get the source code to every job and sequence to you can modify to meet your needs.