what's the recovery objective/target of this database server in question?

A better practice is to print out the result of each alter statement and pipe all output into a log, so easy to find out which procedure is causing error...

I think you can define the shout condition for "NOTOK" in case of the job failure and "LONGRUNNING" if it runs longer than expected. ControlM does keep a log for...

This is absolutely correct. I3 seems to turn on 2861 secretly when it's trying to retrieve the sql stmt.

This traceflag 2861 can be nasty when server is under memory...