3 Replies Latest reply on Aug 7, 2020 7:09 AM by dan.smith@informatica.com

    CDC Condense file Issue

    sagar sawant Active Member

      Hi,

       

      We have set parameter for condense file to  generate after  every 2 mins. But due to sudden migration of source db server It has been observed that there is delay in generation of files. (They are getting generated at random time). Also the file size is  more than usual.

      Also we have maintained sqlora.net to reduce network delay as suggested by Informatica support team.

      But still facing same issue.

      Kindly help me!

       

      ############################# These are the parameter i have configured in my config file###############

      /* Number of minutes to wait between CAPI read cycles in seconds

      NO_DATA_WAIT=0

      /* Number of seconds to wait at the end-of-log for more change data

      NO_DATA_WAIT2=60

       

       

      /* Number of seconds before flushing, or writing, data to the current log file on disk

      /* -1 = No flush, 0 = flush every record, 1 to N flush every N seconds

      /*FILE_FLUSH_VAL=60

      FILE_SWITCH_CRIT=M

      FILE_SWITCH_VAL=2

        • 1. Re: CDC Condense file Issue
          dan.smith@informatica.com Guru

          FILE_SWITCH_VAL doesn't guarantee a file switch.

          It provides an opportunity, depending on what else is going on.

           

          When PWXCCL completes a condense cycle, it checks FILE_SWITCH_CRIT and FILE_SWITCH_VAL.

          If those conditions have been met, then it goes through File Switch logic.

           

          Please not that condense cycles depend on data being delivered from PWX Capture to PWXCCL.  They can take variable amounts of time, depending on how things are going in PWX Capture.

           

          PWX Capture can take varying amounts of time, depending on thow things are going in terms of accessing the source.  If there is a lot of volume on the source DB, that can exceed PWX CDC's ability to capture changes.

           

          Depending on what your source is and other factors, that can vary a lot.

           

          During File Switch logic, several things happen, including maintenance on the CDCT and CND files, based on COND_CDCT_RET_P.

          Another thing that happens is the current CND file is checked.

          If there is anything in it, then it is closed, CDCT is updated, and a new CND file is opened.

          If there isn't any data in the current CND, then it doesn't get closed.  No empty CND files.

           

          If you need more information than this, then I would suggest raising a case with Support.

          1 of 1 people found this helpful
          • 2. Re: CDC Condense file Issue
            sagar sawant Active Member

            Agreed with you dan.smith@informatica.com, even though file size is huge it takes few minutes to read it but what about network delay the data which I'm getting in my DB is generally having delay of n minutes.(Delay in generating condense file) How to Overcome this ?

            Suggested by informatica i have already maintained separate sqlnet.ora and maintained his path in bash profile

             

            Also It has been observed that suddenly it stops writing condense file even though service is up so for this I tried hitting query on source db to check max of record date so everything seems perfect there

            • 3. Re: CDC Condense file Issue
              dan.smith@informatica.com Guru

              I have almost never seen network delay to be a factor.

              You would have to be working with a really small network pipe for that to matter.

               

              Most times, when customers think that they have a "delay" type problem it is actually a capacity problem.

              That is, Oracle is generating more REDO/hour than PWX can consume.I

               

              PWX writes condense files when they are full.

              There isn't any kind of throttle on that.

              If a CND file is full, it gets closed and another one gets opened.

               

              So if you think PWXCCL is closing CND files too slowly, that indicates a problem upstream from that, which means PWX CDC.

               

              PWX CDC processes data as it is available from Oracle.

              If PWX CDC is processing "too slowly", that means that data from Oracle is not available quickly enough.

               

              Most often, that means that you are sourcing from ASM, rather than from disk.

              In those cases,  you may want to look at ASMSTAGING, which is doc'd at PWX 10.4.1, but is actually available as far back as PWX 10.2.0 EBF06.