responses fail when image cache fails

When I accidentally configured glance image cache with a higher limit than the filesystem would support, responses would fail in odd ways. Here's the traceback I saw in glance api logs

2012-02-08 21:20:42 23377 DEBUG [eventlet.wsgi.server] Traceback (most recent call last):
  File "/usr/lib/pymodules/python2.6/eventlet/wsgi.py", line 347, in handle_one_response
    for data in result:
  File "/usr/lib/python2.6/dist-packages/glance/image_cache/__init__.py", line 227, in tee_iter
    cache_file.flush()
  File "/usr/lib/python2.6/contextlib.py", line 34, in __exit__
    self.gen.throw(type, value, traceback)
  File "/usr/lib/python2.6/dist-packages/glance/image_cache/drivers/sqlite.py", line 329, in open_for_write
    yield cache_file
  File "/usr/lib/python2.6/dist-packages/glance/image_cache/__init__.py", line 225, in tee_iter
    cache_file.write(chunk)
IOError: [Errno 28] No space left on device

I understand that one shouldn't misconfigure glance, but it would be nice if a cache failure didn't kill the whole response.