You are viewing a plain text version of this content. The canonical link for it is here.
Posted to general@gump.apache.org by "Adam R. B. Jack" <aj...@trysybase.com> on 2003/10/23 16:58:10 UTC

Issue: dist_ant... timout

Hmm, interesting....

http://lsd.student.utwente.nl/gump/ant/build/build_ant_dist-ant.html

Exit Code: 9 ... means I zapped it after 30 minutes. That said, so did
covalent (after 20 minutes):

http://gump.covalent.com/log/dist-ant.html

BUT: It says BUILD SUCCESSFUL Total time: 36 minutes 22 seconds

Basically even though we stop waiting on it, it does (soon after) build to
completion.

I wrote code (the Nth time, Python wasn't my friend here) to timeout after a
period. It uses a timer thread. Trouble was the thread was a new child PID
on Linux, and when I tried to pkill -KILL -P ${gumpPID} I was killing the
thread & leaving a mess. I used pgrep -P ${gumpPID} -l and manually killed
everything that wasn't 'python', but this has the effect of killing the
shell that runs java to run ant, and not killing the latter two -- since
they are grandchildren of Gumpy, and so forth. Anybody know a way I can kill
a child and those below it, or do I have to write that code? [Sorry, my Unix
is rusty at best, and pre-Linux & toys like pkill/pgrep.]

Q: Ought we allow per-project timeouts? I can code it quite easily.
Something like timeout="N" (in minutes) on the ant/script tag.

regards,

Adam
--
Experience Sybase Technology...
http://www.try.sybase.com