I would suggest looking at what types of notifications they're looking for. When I'm been working on AI jobs, I tend to put failure notifications within the AI jobs and transformations themselves.
For instance, within transformations I use a combination of 'Filter rows' to define when something is not meant to dump information in a file and then send that file through the 'send email' step. I also use the 'error handling' connection (mostly on the AR steps) to write errors to a file and then send that file through email via 'send email' step. This is probably more "error handling" then monitoring, but for me if gives me the ability to be notified on errors while letting the AI job continue through without having to error out. In AI jobs, I use the failure connection to send an email when a specific transformation returns an error to notify me when the transformation didn't complete correctly. The only thing that the above doesn't provide me is in events where an AI job doesn't kicked kicked off via the scheduler. The only times we have seen this within our environment is when the diserver runs out of memory, which we are currently working on putting in monitoring and automation to restart the diserver when it reaches a certain memory threshold. The way I had identified our memory issue is that I had seen the following errors within the ‘\BMC Software\ARSystem\Arserver\Db\arcarte.log on the server: java.lang.OutOfMemoryError: Java heap space java.lang.OutOfMemoryError: GC overhead limit exceeded You can look at tying other monitoring if needed into the arcarte.log file to look for specific events as well. _______________________________________________________________________________ UNSUBSCRIBE or access ARSlist Archives at www.arslist.org "Where the Answers Are, and have been for 20 years"
