build fails without reason, bug in plague?

[Mike McGrath]

Dan Williams wrote:
> On Sat, 2007-08-18 at 19:00 -0500, Mike McGrath wrote:
>   
>> Marek Mahut wrote:
>>     
>>> Hello maintainers,
>>>
>>> Today, I received an error that my build failed, but after deep look on
>>> logs I found that there was no real reason why this build failed [1].
>>> Adel Gadllah told me that it isn't the first time it happens.
>>>
>>> Thoughts?
>>>
>>> [1]
>>> http://buildsys.fedoraproject.org/logs/fedora-4-epel/35730-cvsweb-3.0.5-2.el4/noarch/
>>>
>>>   
>>>       
>> Please add any of these you find to the following ticket.
>>
>> https://hosted.fedoraproject.org/projects/fedora-infrastructure/ticket/125
>>     
>
> The error comes from the _watch_mock() function in builder.py on the
> builders; if somebody could add some logging to the plauge-builder RPM
> and then deploy it, that would help:
>
>     def _watch_mock(self, good_exit, bad_exit):
>         (aux_pid, status) = os.waitpid(self._childpid, os.WNOHANG)
>         status = os.WEXITSTATUS(status)
>         if aux_pid:
> +           self._log("_watch_mock(): PID %d exit status %d\n" % (self._childpid, status))
>             self._childpid = 0
>             if status == 0:
>                 self._done_status = good_exit
>             elif status > 0:
>                 self._done_status = bad_exit
>
>             self._copy_mock_output_to_log()
>             self._start_cleanup()
>
> Dan
>
>
>   
Thanks dan, I've added this to the builders and it appears to be 
working.  All people with this failed sucess please resubmit and let us 
know how it goes.

http://buildsys.fedoraproject.org/logs/fedora-5-epel/35767-nagios-2.9-1.el5/i386/job.log 
<-- working example.

    -Mike