For the basics have a good look at How the DATA Step Works: A Basic Introduction. It is essential learning.   - Jan. ... View more

What you see is because the put _all_ statement is before the input. The datastep halts when the input statement reads end of data in the cards. That's only after you print _all_ to the log.   Also, because the put statement comes before any other statements all variables are missing. At the start of an iteration the PDV is cleared for all non-retained variables.   For illustration you could move the put statement to the botrtom of the datastep after the assigment tof Y.   Hope this clariefies things a bit. - Jan ... View more

Thanks @PaulHomes for confirming my observations. And for putting the thread on track again ;-).   I am still pondering if I will report this to TS as it is not a clear bug. Also there is no immediate impact and we already have a long list of tracks. I am now consider addressing the issue at SGF in the Quad. And in that case we'll have plenty of opportunity to dicsuss the outcome.   Either way I will keep you posted.   Cheers Jan ... View more

Sorry @jakarman but your assumptious about what is being asked here are completey wrong. We are trying to get an answer to one single question: what is the usage of the Output Directory property of the Batch Server. Several things go wrong in your line of answering that result in total derailment of the thread that make me afraid I will not get an answer at all:   1) You assume a different question should be asked and answer the question you think should be asked. Your assumption is wrong and thus your answer. 2) Your answer is lengthy and its language confusing and inconcistent. You lost me at the second sentence. 3) You assume local knowledge from the workfloor ("OB DM IH ...") that we share to be the basis of my question. This assumption is also wrong. Otherwise approach me on that workfloor and not on the forum. 4) You refer to local knowledge that no one will be able to comprehend. Wrong again. 5) You target @Resa based on an incorrect assumption of what I am asking. You also incorrectly assume Resa has some batch thing that is not working. He doesn't, nor have I.   How can answering a simple question go wrong in so many ways?   Another simple question to you: will you please never answer any of my questions?   Regards, Jan. ... View more

Ah yes of course you're right. *Facepalm*. Silly mistake. Nice catch.   - Jan. ... View more

I would suggest a small change to the datastep. The number of observations is now hardcoded. Using NOBS= would make it more flexible:   data brainest; set brainest nobs=n; mle = _SSE_ / n; run; See Example 10: Performing a Function until the Last Observation Is Reached   Regards, - Jan. ... View more

Is this your actual code or just a snippet for illustration? In the datastep you are dsestroying dataset EST. This would be better:   Proc Reg Data=Data1 outest = est tableout; Model y = x; Run; Data est; set est; *<=== read the output estimates from proc reg; MLE = SSE / (_N_);   Also it looks like you may be confused between the number of observations and the row number (_N_). But, again, we may not be be looking at the actual code but just a skeleton.   Kind regards, - Jan. ... View more

Thanks @LinusH. Yes I have. Many times.   The batch server is in fact not even a real server. It is a fancy way of defining the command line that's sent to the OS, in my case via LSF. The command is determined it two stages: first from metadata when deploying (eg from DI Studio) and later by adding the job to a flow. I have noticed that the latter tweaks the log line by adding the names of flow user to the logfile name. I have not added logconfigloc so no log4sas.   In all this I see no sign of adding or handling procedure output via the -print parameter, as the new server metadata property would suggest.   I hope for someone from the forum to shed some light in this. If not I will stalk the developer(s) at SGF 😉   - Jan. ... View more

Hi @JuanS_OCS,   Thanks for chiming in. That does make sense and it is (almost) exactly what I was hoping for. However, the expected does not happen. At least not for me. I have set it up as in the image but a job containing a proc print fails with   ERROR: Insufficient authorization to access /srv/SASConfig/Lev1/SASApp/Options_Libnames.lst.     So SAS tries to stick the .lst file in the root of the batch server config, ignoring the setting in the metadata. Also the command given to LSF to be execute does not contain the metadata setting. I would have expected a -print parameter.   I have taken care of .lst file locations many times, mostly by editing the sasbatch.sh script, but was hoping for things to become more automatic in 9.4M3. So I must be doing something wrong or missing something here. But I don't see it.    - Jan. ... View more