I was having problems with my SAS jobs getting killed whenever my session ended. I came across this thread suggesting use of the nohup command to prevent jobs from shutting down when they receive the hup signal. So, in my .bashrc file, I aliased the sas command as nohup sas so that anytime I run a SAS job, I invoke the nohup command. Sounds good so far, right?

Today I was running a long SAS job and it was randomly killed. I found this error message in the log file:

ERROR: System is Terminating Due to Hang Up Signal (SIGHUP)

How can SAS be responding to the hup signal when I started the job using nohup?

In the thread I cite above, folks discussed using the screen command. Does anyone know if screen can be used remotely (I log in remotely using SSH) or do you have to be right in front of your, umm, screen to use it?

Thanks

screen is a virtual terminal, which means it can be attached and detached from anywere, over ssh, locally, etc.
The problem with screen is that its just that: a terminal, so it does not prevent anything from dying. so you are not guaranteed that program will nto die next time it sees term signal. and btw, when you use nohup the correct formatting is 'nohup [command] &' -- & sign makes sure that the thread is detached and goes into the background so your closure of ssh session does not kill nohup execution.

Yes, that's the way I'm using nohup. I always run my sas jobs in background (with &), so by aliasing sas to nohup sas, everytime I command sas job.sas &, it executes as nohup sas job.sas &. Thanks for checking though.

A login shell does not read .bashrc. Perhaps that is the cause of the problem. Might I suggest that you create a .profile and have a line in there to source your .bashrc.

Perhaps, but the session was recognizing all the other aliases I had set up in the .bashrc. I'll dig around to learn how to set up a .profile and see if that fixes the problem. Thanks.