----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/58422/#review172611 -----------------------------------------------------------
distro/src/bin/atlas_stop.py Line 71 (original), 74 (patched) <https://reviews.apache.org/r/58422/#comment245698> The fix looks good. A couple of things you could change: - you could correct the seconds seconds in the message you altered - you could add comments to detail why this was necessary - you could keep the signal name in the message and have 2 more specific messages. These are my suggestions - it is up to you whether you make these changes as they are all quite minor. - David Radley On April 13, 2017, 4:28 p.m., Graham Wallis wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/58422/ > ----------------------------------------------------------- > > (Updated April 13, 2017, 4:28 p.m.) > > > Review request for atlas and David Radley. > > > Repository: atlas > > > Description > ------- > > On Windows there is no SIGKILL in the python signal library, so attempting to > import SIGKILL causes a fatal error. > This change introduces a platform switch that avoids trying to import SIGKILL > and if the 30sec timeout expires > it avoids attempting to use SIGKILL, instead using SIGTERM as the os.kill() > function on Windows will terminate > the process and should be as severe as a SIGKILL (kill -9) on a *nix system. > > > Diffs > ----- > > distro/src/bin/atlas_stop.py a25d25aee599e7cc9ca3caaff8ff7f11b7e0c789 > > > Diff: https://reviews.apache.org/r/58422/diff/1/ > > > Testing > ------- > > Functional testing manually verified > > > Thanks, > > Graham Wallis > >
