On Mon, Sep 10, 2012 at 12:18:46PM -0400, Kevin A. McGrail wrote: > On 9/8/2012 7:49 PM, Marc Andre Selig wrote:
> >Yesterday was the second time the weekly --net check has not run for me. > >No log files were created at all. > > > >Trying to debug things, I just ran the mass-check command individually > >(with the options as assembled by auto-mass-check.sh), but I got no output > >at all, with no files being created either, and mass-check exited with > >a return code of 141 after maybe two seconds of waiting. I still don't receive any --net results. A manual invocation of mass-check fails the first time around, then succeeds when called again immediately. Which makes the error kind of difficult to debug. ;) > I don't know what a return code of 141 means and I didn't see > anything obvious. > > Can you run an strace on the command next time and perhaps that will > shed more light? strace shows that mass-check dies after receiving a SIGPIPE signal (well, 141-128=13=SIGPIPE). If I read things correctly, a cloudmark (razor) server resets the connection, then mass-check still tries to write to the corresponding fd, resulting in a broken pipe error and signal. Sorry, but there's not much time left for me now, so I can't debug this any further at the moment. I might be misinterpreting things anyway, as this is not my primary line of work. I've put the strace output online at <http://www.sedacon.com/private/masscheck-121004.strace.bz2>, in case anybody else wants to have a look. Regards, Marc
