[GE users] SGE questions

Andreas.Haas at Sun.COM Andreas.Haas at Sun.COM
Thu Mar 1 17:39:51 GMT 2007

Hi John,

On Thu, 1 Mar 2007, john.li at mindspeed.com wrote:

> Great,  thanks.  Currently I have the source code changed to 022  in the
> files suggested.

Changed to(?) 022 ... it is 022 right now.

> I had to change a few other files to get the stdout and stderr too 002.

Well, more that those shouldn't be needed in addtion

    daemons/execd/exec_job.c:      umask(022);
    daemons/execd/execd.c:   umask(022);

the others here do not impact the job start

    daemons/qmaster/setup_qmaster.c:   umask(022);
    daemons/schedd/sge_schedd.c:   umask(022);
    daemons/shadowd/shadowd.c:   umask(022);

> But I'll certainly try the prolog suggestion.

That works at all events.

> Yes, I'll be happy to verify the JOB_UMASK when it is available.  Is it in
> v61beta?

We must see. First at all events is agree on the requiments.

> While you are at it, if I may, please provide something like STDOUT_UMASK
> as well.   The reason this umask becomes a very important requirement is
> that we allow
> all group access at a project level.   So project team member can easily
> work on other people's
> work.   Even a SGE output file, without 002, it may stop other group
> members' work flow.

Nevertheless I don't see a need for STDOUT_UMASK + STDERR_UMASK.
The JOB_UMASK parameter would impact both of them. IMO adding 
two parameters is classic overengineering ;-)

> Come to think of , why can we not inherit the umask from user's shell
> environment?

Because stdout/sterr are opend before the shell is even started.


To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net

More information about the gridengine-users mailing list