Description of problem:
There's a bug where the _fail method is called from two different places.  I noticed this when testing patches for https://bugzilla.redhat.com/show_bug.cgi?id=1440383  Originally I thought maybe that patchset introduced the problem but then after reverting to a previous build that didn't have any of those patches I still had the problem.  So then I put in traceback.print_stack() at the top of the _fail() method and found the following two places _fail is called:

==> /var/log/httpd/error_log <==
  File "/usr/lib/python2.6/site-packages/retrace/retrace_worker.py", line 743, in start_vmcore
    self._fail()
  File "/usr/lib/python2.6/site-packages/retrace/retrace_worker.py", line 101, in _fail
    traceback.print_stack()
  File "/usr/bin/retrace-server-worker", line 67, in <module>
    worker.start(kernelver=kernelver, arch=cmdline.arch)
  File "/usr/lib/python2.6/site-packages/retrace/retrace_worker.py", line 955, in start
    self._fail()
  File "/usr/lib/python2.6/site-packages/retrace/retrace_worker.py", line 101, in _fail
    traceback.print_stack()

Here's the code for the first call:
    544     def start_vmcore(self, custom_kernelver=None):
    545         self.hook_start()
    546 
    547         task = self.task
...
    581         if task.use_mock(kernelver):
...
    734         else:
    735             try:
    736                 self.hook_pre_prepare_debuginfo()
    737                 crash_cmd = task.get_crash_cmd().split()
    738                 vmlinux = task.prepare_debuginfo(vmcore, kernelver=kernelver, crash_cmd=crash_cmd)
    739                 task.set_crash_cmd(' '.join(crash_cmd))
    740                 self.hook_post_prepare_debuginfo()
    741             except Exception as ex:
    742                 log_error("prepare_debuginfo failed: %s" % str(ex))
    743                 self._fail()   <--------------- 1st call to _fail


code for the second:
    902     def start(self, kernelver=None, arch=None):
...
    947             if tasktype in [TASK_RETRACE, TASK_DEBUG, TASK_RETRACE_INTERACTIVE]:
    948                 self.start_retrace(custom_arch=arch)
    949             elif tasktype in [TASK_VMCORE, TASK_VMCORE_INTERACTIVE]:
    950                 self.start_vmcore(custom_kernelver=kernelver)  <---------- raises Exception on 741 above
    951             else:
    952                 raise Exception("Unsupported task type")
    953         except Exception as ex:   <----------- second Exception raised
    954             log_error(str(ex))
    955             self._fail()   <------------- 2nd call to _fail


Version-Release number of selected component (if applicable):
retrace-server-1.17.0.111+gb07f-1.el6.noarch
rpm -q retrace-server --changelog | head
* Wed Aug 16 2017 Dave Wysochanski <dwysocha> 1.17.0.111+gb07f-1
- Build at upstream b07f88d Merge pull request #159 from DaveWysochanskiRH/rhbz1481433-crash-directory-unreadable

* Thu Mar 30 2017 Matej Marusak <mmarusak> 1.17.0-1
- Enable creating releases with makefile
- Introduce gen-version
- Do not use fedorahosted.org as source
- Include md5sum of original archive in summary page
- Do not try to get default time when ftptask
- Fix character escape typo

How reproducible:
everytime

Steps to Reproduce:
1. submit a vmcore that will fail to retrace due to some reason (failing debuginfo, etc)

Actual results:
_fail called twice

Expected results:
_fail only called once

Additional info:
There's also an odd 'None' log line in the retrace-log file.  Example:
   2018-01-17 08:55:15 prepare_debuginfo failed: Unable to find debuginfo package and no cached vmlinux file
    2018-01-17 08:55:15 Sending e-mail to dwysocha
   2018-01-17 08:55:15 None    <--------------- Looks like the "log_error" from line 954 above
    2018-01-17 08:55:15 Sending e-mail to dwysocha  <--- duplicate