This only happened under heavy load (peak processing + someone left a server snapshot on). For 2 days in a row, 2 packages which are part of 2 separate SQL server agent multi step jobs did not create an extract and the next step in the job started (this is not the trivial error where you forget to set "On failure Quit the job reporting failure"). In the job history, I see the steps reported DTSER_SUCCESS(0). All our packages run in a framework, all auditing info looks like it's supposed to look (step never ran, set to "ready" (not " complete")). After the snapshot was stopped, processing looks normal again (all run times back to normal, both packages creating their extracts).
I wonder if something isn't set correctly within (all) the packages. Does FailParentOnFailure have to be set to 'true' for all components?
Thanks!