4 Replies Latest reply on Oct 19, 2019 12:17 PM by Mark Francome

    concurrent job failures in Informatica

    Bruce S
      Share This:

      Around 1000 jobs are running in a few minutes in Informatica server. Informatica has a master node and about 8 slave nodes. And these jobs running in the slave nodes. If the connection between Master and slave stopped because of a network glitch, all the 1000 jobs failing. There is a quantitative resource set up around 100. Now, if the jobs started failing, we are manually changing the resource number to 0(but it is too late before someone go and change to 0 and all the jobs failed by that time). is there any way we can avoid this excessive job failures / manual task to change the quantitative resource to 0?

        • 1. Re: concurrent job failures in Informatica
          Bhanu Prakash Badiginchala

          I'm not familiar with Informatica jobs, so not sure if there is a better way.

           

          If there is specific OSCOMPSTAT for this network failure or if this is the only failure expected (OSCOMPSTAT not 0), this might work:

           

          Have a common IN condition to all of them. In the job's actions, on required OSCOMPSTAT, remove the condition. Ofcourse, that would mean the condition needs to be added manually to resume the jobs.

          1 of 1 people found this helpful
          • 2. Re: concurrent job failures in Informatica
            Mark Francome

            You can automatically update the quantitative resource via the ecaqrtab command line utility -

             

            ecaqrtab update <QR Name> <QR Quantity>

             

            You could have an "on demand only" job with that command and force it in whenever an Informatica job fails.

             

             

            1 of 1 people found this helpful
            • 3. Re: concurrent job failures in Informatica
              Bruce S

              Is there anyway we can count the number of job failures on Regular folder(NOT SMART folder) and trigger a job based on that? Am thinking that if this folder cross 20 failed jobs , then trigger a job to make the quantitative resource to 0.

              1 of 1 people found this helpful
              • 4. Re: concurrent job failures in Informatica
                Mark Francome

                You could create a cyclic job that runs a job to query the number of failures on those jobs (via ctmpsm) and then uses ecaqrtab to set the number to zero.

                 

                Alternatively use a counter to increment +1 for every failure. Echo out the count in the prev command field and if it exceeds 20 then get the command to echo text that you can then match on in a "force job" step in post-processing. The only thing to remember with this method is that you will have to manage the counter, i.e. set it to zero when the daily batch has ended.

                1 of 1 people found this helpful