topic Studio Linux output dataset permissions vs. Batch dataset permissions in SAS Studio

Studio Linux output dataset permissions vs. Batch dataset permissions

epniv — Thu, 18 Sep 2025 13:10:13 GMT

Hi All,

Something odd with output datasets, or other output file types, with our Linux Studio implementation.

When running batch SAS files on our server the output file permission are "rw-rw-r--". This is for datasets or XLSX files, etc.

However, when running the exact same code in Studio the output file permission are "rw-r--r--". This means that the datasets are not "write" shareable within the same group of staff.

I've made sure that all staff have "umask 002" in their .bash_profile file. Which should, and does, lead to creating rw-rw-r-- file permissions, like in batch processing.

Does anyone know if there is a configuration change that can be made for Studio to default to rw-rw-r--?

We do know that the permissions can be set via an X command within Studio code (or other programmatic methods) but with many different datasets created via some macros and names that are not always constant, this is not preferrable.

Thanks!

Re: Studio Linux output dataset permissions vs. Batch dataset permissions

Quentin — Thu, 18 Sep 2025 13:57:35 GMT

Sounds like this is Studio connected to a SAS 9 linux server, not Viya, is that correct?

Just curious, is there a difference in the owner of the output file when the job is run in batch vs via Studio?

I think it's pretty common for scheduled batch jobs to be run under a different account.

It links to:
https://sas.service-now.com/csm/en/setting-umask-and-ulimit-values-for-sas-sessions-on-unix-and-linux?id=kb_article_view&sysparm_article=KB0036220

Re: Studio Linux output dataset permissions vs. Batch dataset permissions

epniv — Thu, 18 Sep 2025 14:26:00 GMT

Hi Quentin,

Thanks for the questions, comments and links.

This is not a Viya system. It is Studio connected to a SAS 9 linux server, AnalyticsPro installation. Just one server, no metaserver.

Yes, these are the same individual users generating the output in Studio vs. batch. So, not scheduled batch jobs.

The KB0036220 link looks promising but I'll need to do some testing on that or wait for some more comments from others who may have solved this.

Re: Studio Linux output dataset permissions vs. Batch dataset permissions

Kurt_Bremser — Thu, 18 Sep 2025 17:29:53 GMT

Batch jobs "inherit" permissions from the scheduler process. Add a umask command to the sasbatch_usermods.sh file of your application server context.

Re: Studio Linux output dataset permissions vs. Batch dataset permissions

quickbluefish — Fri, 19 Sep 2025 00:01:50 GMT

A bit off-topic, but I've taken to connecting to the SAS Studio server via Putty and just doing everything - developing and running code, file management, etc. - via the command line (and various bash functions / aliases stored in ~/.bashrc). I open this connection with WinSCP (which subsequently launches Putty), and this gives the added benefit of being able to download all files at once - something that there's still not a simple, built-in way to do with the Studio interface.

Anyway, you might try that if there's no more direct way - just make a simple function that uses chmod to convert permissions on all *.sas7b* files, for example.

Re: Studio Linux output dataset permissions vs. Batch dataset permissions

epniv — Mon, 22 Sep 2025 10:47:23 GMT

Hi All,

Sorry if this is out of the norm for a response.

Quentin's suggestion in the second link referencing KB0036220 had the answer. By adding "umask 002" to my "workspaceserver_usermods.sh" config file did the trick.