Joined: 26 Apr 2004 Posts: 4650 Location: Raleigh, NC, USA
I'm looking for some ideas for industry-best practices with regards to generating and assigning problem tickets in a Problem Management System for batch job abends/failures, and for started task abends.
Some things I'm looking for:
- What facility creates and assigns problem tickets?
- What criteria are used to determine IF a ticket is generated and to whom it is assigned?
- Are problem tickets assigned different levels of severity depending on the criticality of the batch application that has failed?
- Do you create multiple tickets for multiple failures for the same job/task?
- How is the assignment process managed? How is it audited?
- Does the creation of a problem ticket drive an automated contact call-out to the support team?
- Is the process consistent across mainframe/non-mainframe platforms?
Joined: 20 Oct 2006 Posts: 6970 Location: porcelain throne
I think your question and comments scared everyone. We know that most of the members that work at out-sourcing have this sort of thing.
Maybe better phrased would be:
What happens at your shop when one of your programs has problems in production?
Most shops I have worked at have operations turn-over documentation. Included is the old what-to-do-when-it-falls-over narrative.
Been in europe since mid-90's and here the automation consists of an automated e-mail to the bureaucracy that deals with bad-things sent by the scheduler when an unacceptable JOB return code event occurs. This group then badgers the offending programmer with paperwork until the group has determined that they have collected enough forms to satisfy the severity of the offense. Being germany, I am sure that have some system for dealing with the programmer responses (other than estimating the number of trees involved); but the net effect is to insure in the future that as a programmer, your name is never on a responsible-for list.
Most of the determination work is accomplished by the scheduler, based on the sophistication of the scheduling system. Most of the criteria you mentioned would be interpreted into parameters.
with the advent of web development, I notice most vendors (IBM included) now sell ProblemManagementSystems dedicated to finding problems before they happen. e.g. work-flow analysis, hardware configuration vs expected workload. There is more, but I think they have stopped trying to sell management systems that more or less only noted that the-barn-door-was-left-open-last-night and have found more lucrative markets in the pro-active type systems.
I know I did not help you much, but I did increase the replies on your thread and maybe my attempts at humor will induce others to write of their experiences - hopefully beneficial to your quest.