right now we reset our mock environment and usually re-install the same packages over and over for each build. we should save the list of packages that have been installed into the mock environment somewhere, and if the build we're doing is using the same set we can skip resetting the environment. we should still run the install step, which will be a noop in most cases, but allows us to update packages in the yum repo without having to clobber the mock environment on all the slaves.

I took a quick stab at this and ran into problems with subsequent mock --copyin steps: 20:16:21 INFO - INFO: copying /home/cltbld/.ssh to /builds/mock_mozilla/mozilla-centos6-i386/root/home/mock_mozilla/.ssh 20:16:21 INFO - Traceback (most recent call last): 20:16:21 INFO - File "/usr/sbin/mock_mozilla", line 862, in <module> 20:16:21 INFO - main(retParams) 20:16:21 INFO - File "/usr/sbin/mock_mozilla", line 821, in main 20:16:21 INFO - shutil.copytree(src, dest) 20:16:21 INFO - File "/usr/lib64/python2.6/shutil.py", line 142, in copytree 20:16:21 INFO - os.makedirs(dst) 20:16:21 INFO - File "/usr/lib64/python2.6/os.py", line 157, in makedirs 20:16:21 INFO - mkdir(name, mode) 20:16:21 INFO - OSError: [Errno 17] File exists: '/builds/mock_mozilla/mozilla-centos6-i386/root/home/mock_mozilla/.ssh' 20:16:21 ERROR - Return code: 1 20:16:21 FATAL - Halting on failure while running ['mock_mozilla', '-r', u'mozilla-centos6-i386', '--copyin', '--unpriv', u'/home/cltbld/.ssh', u'/home/mock_mozilla/.ssh'] 20:16:21 FATAL - Exiting 1 Maybe we should be fixing mock to not error out in this case?