Anytime 🙂 ... View more

Hi, thank you so much for the tips.   How could accept multiples propositions that work as a solution? When I accept one, I can't no longer accept other solution  as a solution too. ... View more

You could use data step UPDATE statement instead. Let's create some test datasets. data have; do id=1 to 5; output; end; run; data test1; input id value; cards; 1 . 2 4 ; data test2; input id value; cards; 1 3 2 5 3 . ; data test3; input id value; cards; 1 6 2 7 3 8 4 . ; Now stack the "test" data together into a transaction dataset. Use the INDSNAME= option of the SET statement to find the name of that dataset that is supplying each record. data trans; set test1-test3 indsname=indsname; by id; where not missing(value); dsname=indsname; run; Now apply these transactions to you original dataset.  Keep track of whether or not a non missing value has been found and only write that observation instead of the normal UPDATE behavior of just writing the final observation for the group. data want; update have trans; by id; if first.id then found=0; if not missing(value) and not found then do; found+1; output; end; if last.id and not found then output; run; Results Obs id value dsname found 1 1 3 WORK.TEST2 1 2 2 4 WORK.TEST1 1 3 2 5 WORK.TEST2 1 4 3 . 0 5 4 . 0 6 5 . 0 ... View more

Thanks! ... View more

No, there isn't a quick way to kill all jobs. You can kill all jobs by creating a SAS program that kills SAS processes (see link below for Unix server) or use O/S commands/scripts. Please be careful when doing this as you will be killing all jobs, including batch jobs that have been submitted to run overnight. If you decide to use a SAS program instead of O/S scripts, once you create and test your SAS program then you can deploy that job from within Management console and create a flow and schedule it to run every day.   The SAS Terminator: An Ultimate Killer of Unix Processes (lexjansen.com)      ... View more