This is similar to our old friend https://bugzilla.redhat.com/show_bug.cgi?id=1315541 , but doesn't look to have the same cause.

The Fedora-26-20170309.n.0 Workstation live image compose failed:

https://koji.fedoraproject.org/koji/taskinfo?taskID=18278005

The problem is failure to umount /dev , again. anaconda's log - https://kojipkgs.fedoraproject.org//work/tasks/8005/18278005/anaconda-program.log - shows this:

12:53:24,606 INFO program: Running... umount /mnt/sysimage/dev
12:53:24,620 INFO program: stderr:
12:53:24,620 INFO program: b'umount: /mnt/sysimage/dev: target is busy'
12:53:24,621 INFO program: b'        (In some cases useful info about processes that'
12:53:24,621 INFO program: b'         use the device is found by lsof(8) or fuser(1).)'
12:53:24,622 DEBUG program: Return code: 32
12:53:24,622 INFO program: Running... lsof /mnt/sysimage/dev
12:53:24,796 INFO program: stdout:

then the lsof output lists just about every running process, starting with systemd, indicating that possibly systemd is the culprit. Then, as usual, livemedia-creator tries to clean up the mount that anaconda left in place, and fails, and this is counted as a fatal error. See https://kojipkgs.fedoraproject.org//work/tasks/8005/18278005/root.log :

DEBUG util.py:435:  2017-03-09 12:53:25,067: Error in atexit._run_exitfuncs:
DEBUG util.py:435:  2017-03-09 12:53:25,068: Traceback (most recent call last):
DEBUG util.py:435:  2017-03-09 12:53:25,068: File "/usr/lib/python3.6/site-packages/blivet/threads.py", line 45, in run_with_lock
DEBUG util.py:435:  2017-03-09 12:53:25,068: return m(*args, **kwargs)
DEBUG util.py:435:  2017-03-09 12:53:25,068: File "/usr/lib/python3.6/site-packages/blivet/blivet.py", line 1470, in umount_filesystems
DEBUG util.py:435:  2017-03-09 12:53:25,068: self.fsset.umount_filesystems(swapoff=swapoff)
DEBUG util.py:435:  2017-03-09 12:53:25,069: File "/usr/lib/python3.6/site-packages/blivet/osinstall.py", line 664, in umount_filesystems
DEBUG util.py:435:  2017-03-09 12:53:25,069: device.format.teardown()
DEBUG util.py:435:  2017-03-09 12:53:25,069: File "/usr/lib/python3.6/site-packages/blivet/threads.py", line 45, in run_with_lock
DEBUG util.py:435:  2017-03-09 12:53:25,069: return m(*args, **kwargs)
DEBUG util.py:435:  2017-03-09 12:53:25,069: File "/usr/lib/python3.6/site-packages/blivet/formats/__init__.py", line 645, in teardown
DEBUG util.py:435:  2017-03-09 12:53:25,070: self._teardown(**kwargs)
DEBUG util.py:435:  2017-03-09 12:53:25,070: File "/usr/lib/python3.6/site-packages/blivet/threads.py", line 45, in run_with_lock
DEBUG util.py:435:  2017-03-09 12:53:25,070: return m(*args, **kwargs)
DEBUG util.py:435:  2017-03-09 12:53:25,070: File "/usr/lib/python3.6/site-packages/blivet/formats/fs.py", line 560, in _teardown
DEBUG util.py:435:  2017-03-09 12:53:25,071: raise FSError("umount of %s failed (%d)" % (mountpoint, rc))
DEBUG util.py:435:  2017-03-09 12:53:25,071: blivet.errors.FSError: umount of /mnt/sysimage/dev failed (32)
DEBUG util.py:435:  2017-03-09 12:53:59,498: Shutting down log processing
DEBUG util.py:435:  2017-03-09 12:53:59,522: failed to unmount /mnt/sysimage/dev. retrying (1/3)...
DEBUG util.py:435:  2017-03-09 12:54:00,606: failed to unmount /mnt/sysimage/dev. retrying (2/3)...
DEBUG util.py:435:  2017-03-09 12:54:01,676: Cleanup of /mnt/sysimage/dev failed. See program.log for details
DEBUG util.py:435:  2017-03-09 12:54:01,683: failed to unmount /mnt/sysimage. retrying (1/3)...
DEBUG util.py:435:  2017-03-09 12:54:02,740: failed to unmount /mnt/sysimage. retrying (2/3)...
DEBUG util.py:435:  2017-03-09 12:54:03,799: Cleanup of /mnt/sysimage failed. See program.log for details
DEBUG util.py:435:  2017-03-09 12:54:03,800: Install failed: novirt_install cleanup of anaconda mounts failed.
DEBUG util.py:435:  2017-03-09 12:54:03,801: Removing bad disk image
DEBUG util.py:435:  2017-03-09 12:54:03,801: ERROR: Image creation failed: novirt_install cleanup of anaconda mounts failed.
DEBUG util.py:573:  Child return code was: 1

We need to investigate a bit more to figure out if this is a consistent error, or if it's another lottery situation, or what, but it's at least a potential F26 Alpha blocker, so proposing it as such.