[Mailman-Developers] fwd: Cron <mailman@myhost> /usr/bin/python /home/mailman/cron/senddigests (fwd)

[Jerry Adlersfluegel]

Harald Meland wrote:
> 
> [Jerry Adlersfluegel]
> 
> > Traceback (innermost last):
> >   File "/home/mailman/cron/senddigests", line 37, in ?
> >     main()
> >   File "/home/mailman/cron/senddigests", line 34, in main
> >     list.SendDigestIfAny()
> >   File "/home/mailman/Mailman/Digester.py", line 194, in SendDigestIfAny
> >     self.SendDigestOnSize(0)
> >   File "/home/mailman/Mailman/Digester.py", line 206, in SendDigestOnSize
> >     self.SendDigest()
> >   File "/home/mailman/Mailman/Digester.py", line 290, in SendDigest
> >     self.DeliverToList(d.Present(mime=0),
> >   File "/home/mailman/Mailman/Deliverer.py", line 172, in DeliverToList
> >     status = cmdproc.close()
> > IOError: (10, 'No child processes')
> >
> >
> > Any ideas?
> 
> Yup.  I believe this is a known problem with at least some RedHat cron
> daemons -- they run their jobs with some (rather vital, like SIGCHLD)
> signals set up to be ignored.
> 
> Workaround: I think that adding something like this to the top of any
> script that is to be run from cron would work (as long as it is
> SIGCHLD that is being incorrectly ignored, anyway):
> 
> import signal
> signal.signal(signal.SIGCHLD, signal.SIG_DFL)
> 
> If this workaround does the job, it would IMHO be less obtrusive than
> the workaround currently in CVS.
> --
> Harald

I have been paying attention, and that error report does not get sent
every day. I think it only went out two or three times this week. I did
not get that error today or yesterday. 

I haven't applied anything to the 1.0b7 installation. I'll pay attention
to what happens.

Thanks!