autobuild ctdb cleanup
amitay at gmail.com
Tue Jul 8 00:57:31 MDT 2014
On Tue, Jul 8, 2014 at 9:25 AM, Stefan (metze) Metzmacher <metze at samba.org>
> Hi Amitay,
> >> I noticed that there's a lot of ctdb and ctdb_reco processes
> >> left after an autobuild (maybe only failed once).
> >> Could you take a look at that and may implement something like
> >> the smbd_stdin_handler in source3/smbd/server.c.
> >> So that ctdb can be started in foreground and exit when the parent went
> >> away.
> >> Thanks!
> >> metze
> > Hi metze,
> > There is already logic in ctdb to clean up child processes if the main
> > daemon goes away. The problem here is of shutting down the local ctdb
> > daemons when autobuild.py terminates abnormally.
> Samba has the same cleanup method for both cases...
> > Here's a patch that shuts down the local daemons started for test if the
> > run_tests.sh script is sent SIGINT or SIGTERM. I have tested it by
> > autobuild.py and then pressing Ctrl-C. This sends SIGTERM to run_tests.sh
> > and shuts down the local daemons. So as long as autobuild.py terminates
> > the child processes, this would work.
> Lets give it a try and please push it with my review.
Pushed to autobuild with fixes to waf build. Since I did not check the
output of RUN_COMMAND(), CTDB test failures were not failing autobuild.
More information about the samba-technical