swift team mailing list archive

[Pete Zaitcev <zaitcev@xxxxxxxxxx>]

Dear All:

I am setting up a Swift instance in a chroot on RHEL 6, using SAIO as
a starting point. When running .unittests, the following happens:

ERROR: test_update_once_single_threaded (test.unit.obj.test_updater.TestObjectUp
dater)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/q/zaitcev/hail/swift-20100823/test/unit/obj/test_updater.py", line 126,
 in test_update_once_single_threaded
    raise err
RuntimeError: Second simultaneous read on fileno 184 detected.  Unless you reall
y know what you're doing, make sure that only one greenthread can read any parti
cular socket.  Consider using a pools.Pool. If you do know what you're doing and
 want to disable this error, call eventlet.debug.hub_multiple_reader_prevention(
False)

What's suspicious, everything works fine if I:
 - run a smaller test collection, e.g. for just one test:

nosetests test/unit/obj/test_updater.py:TestObjectUpdater.test_update_once_single_threaded -exe --with-
coverage --cover-package swift --cover-erase

 - run the same .unittest on some other computer (also RHEL 6)

So I suspect the test is not entirely correct and produces a race
that ends tripping self-checks in greenlet.

Does anyone have any tips as to where to look for problems?

-- Pete