Re: [HTCondor-users] output files and umask

[Michael Di Domenico <mdidomenico4@xxxxxxxxx>]

this umask issue has reared it's head again.  can anyone tell me what
the repercussion to changing these umask's in the source from 022 to
027.  We don't permit (by policy) world readable files for various
group owned files.  Unfortunately condor is resetting the umask within
the users jobs being run under dagman.  i'd like to alter the source
code, but i'm not sure i have a full grasp on what the resulting
change would be internally to condor.

[~/Downloads/condor-8.8.3]$ find . -type f -exec grep -H umask {} \; | grep 022
./src/html/build_doc++_index:umask 022;
./src/condor_scripts/make_special_static_tar:umask(022);
./src/condor_scripts/make_final_tarballs:umask(022);
./src/condor_scripts/make_nmi_final_tarballs:umask(022);
./src/condor_scripts/condor_configure:umask 0022;
 # default file creation permissions
./src/condor_shadow.std/pseudo_ops.cpp: omask = umask( 022 );
./src/condor_shadow.std/pseudo_ops.cpp:                 omask = umask( 022 );
./src/condor_shadow.std/pseudo_ops.cpp:                 omask = umask( 022 );
./src/condor_shadow.std/pseudo_ops.cpp: omask = umask( 022 );
./src/condor_shadow.std/pseudo_ops.cpp: omask = umask( 022 );
./src/condor_utils/dprintf.cpp: old_umask = umask( 022 );
./src/condor_utils/email.cpp:   prev_umask = umask(022);
./src/condor_utils/email.cpp:   prev_umask = umask(022);
./src/condor_daemon_core.V6/daemon_core_main.cpp:       umask( 022 );

On Tue, Oct 16, 2018 at 11:36 AM Michael Di Domenico
<mdidomenico4@xxxxxxxxx> wrote:
>
> On Tue, Oct 16, 2018 at 11:18 AM Michael Di Domenico
> <mdidomenico4@xxxxxxxxx> wrote:
> > this still seems to only happen when dagman submits the jobs rather
> > then the user with condor_submit.  i hope that narrows the code path a
> > bit and might be able to have someone tell where i can fix the code
>
> i partially take this part back.  with skip_filechecks = true the
> files are 0022 instead of 0027 when a user does condor_submit instead
> of a dag, if i revert skip_filechecks then i get the behavior as
> described