0

My company is running Autosys r11.1 SP1, as our enterprise job scheduler, and I find the product to serve its purpose fine. It has a reputation in house at our company as being "complex" and "not easy to use"; I maintain that as a cross-platform enterprise job scheduler, it's certainly going to be complex, and the of course it will take some time and dedication to master the administration of such a system.

I am not part of the team responsible for administration, however my team is the heaviest user of the product as I run our data warehouse team, and I am trying to prepare some counter arguments against what I believe to be false claims about limitations of the product. I admittedly know that "Autosys" is a suite of software, but I am by no means an expert. I believe that separate from the actual job scheduler, there is an alerts engine, and also a Workload Control Center, of which we have all three pieces installed.

Currently, if an Autosys job hits Max Run Alarm status, an email alert is generated to our help desk and they are able to take the appropriate action. This, from my understanding of Autosys' internal data model is an "Event" which can happen to a job.

This is different from the various statues which a job can belong to, one at a time, of which I know of;

  1. Activated
  2. Inactive
  3. Starting
  4. Running
  5. Success
  6. Failure
  7. On Hold
  8. On Ice
  9. Late to Start
  10. Machine Pending
  11. Terminated

In addition to the alerts when a job encounters the Max Run event, our help desk also receives an email alert when a job hits a Failure status or Machine Pending status.

I am being told that it is unable to send out an alert if a job hits the Terminated Status? I just don't believe this.

I am also told that there is no way to filter the job name before sending any sort of alert. We currently do not have a true development instance of Autosys, so we use naming conventions to differentiate production versus UAT or Test, and as of now email alerts are generated for all, and we face a constant battle with our help desk trying to get them to understand we don't need tickets created for non-production jobs.

Any guidance or education on the true capabilities of this product would be greatly appreciated!

Chris

Chris Corbin
  • 49
  • 1
  • 6

2 Answers2

0

As I have been a part of autosys for almost 3 years, according to my knowledge there is no alert generated for the job if the job goes to Terminated(TE) status. You get alerts for chase, max run and failed job. If you make the attribute alarm_if_fail: 0 you will not get the alert for the job if it failed.

A suggestion for the tickets which are created by your servicedesk team for non production jobs: you can make slight changes in the job to make the service desk guys realize that this is a production job by making changes in the job name. For example, like this: us_sap_morning_data_p_box. Here, "P" can help you find the production jobs. Similarly for the cmd and fw jobs you can use this: us_sap_morning_data_p_cmd, or us_sap_morning_data_p_fw.

I am not sure how far this helps you.

pjmorse
  • 9,204
  • 9
  • 54
  • 124
som
  • 1
0

One way to generate alerts based on TE events, specifically cantered around boxes which don't get the chance to execute any code within return code check blocks (or traps etc) is to create an alert job. An alert job that looks at the condition of the box (or specific job) would be a command job that sends an email.

The condition would look something like

condition: f(name_of_box_or_job) || t(name_of_box_or_job)
FDinoff
  • 30,689
  • 5
  • 75
  • 96