Open Stack

On 10/26/2012 01:49 PM, Adam Young wrote:
> On 10/26/2012 10:52 AM, Dolph Mathews wrote:
>> Pending feedback from Adam and a bug report, here's a potential fix: 
>> https://review.openstack.org/#/c/14865/
>
> Yes, please test that out.  I would not expect that the greenthreads 
> would intentionall share a file descriptor but there is always the 
> possibility of a bug in the Eventlet layer.
>
> Please file this as a bug.
>
>
>>
>> -Dolph
>>
>>
>> On Thu, Oct 25, 2012 at 7:49 PM, Dolph Mathews 
>> <dolph.mathews at gmail.com <mailto:dolph.mathews at gmail.com>> wrote:
>>
>>     Any details on what file is being read twice? (I doubt this is a
>>     quantum issue -- more likely to be keystone)
>>
>>     I'm guessing the culprit is that the auth_token middleware
>>     reads/writes a PKI token revocation list to disk... although I'm
>>     not sure why it can't be stored in memory -- Adam?
>>
>>     -Dolph
>>
>>
>>     On Thu, Oct 25, 2012 at 5:16 PM, Shiv Haris <sharis at brocade.com
>>     <mailto:sharis at brocade.com>> wrote:
>>
>>         Hi Quantum team,
>>
>>         Lately I have come across a new error using quantum from the
>>         "master". Never saw this before. It appears I may have
>>         updated some package and at this time I have no idea which
>>         one to back off. This may be related to some python package.
>>
>>         Any ideas?
>>
>>         2012-10-25 15:11:12 ERROR [keystone.middleware.auth_token]
>>         HTTP connection exception: Second simultaneous read on fileno
>>         9 detected.  Unless you really know what you're doing, make
>>         sure that only one greenthread can read any particular
>>         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)
>>

It is also possible that the problem is memcached and not the file 
read.  This problem was reported before due to memcache, and the 
solution was to monkey_patch thread.

modify
quantum/wsgi.py:28:eventlet.patcher.monkey_patch(all=False, socket=True, 
thread=True)


in the keystone-all executable, we have this code:

     monkeypatch_thread = not CONF._cli_values['standard_threads']
    eventlet.patcher.monkey_patch(all=False, socket=True, time=True,
                                   thread=monkeypatch_thread)



The ability to turn off monkeypatching of thread is valuable, as it 
allows debugging.  I would  move the code that performs the monkey patch 
in quantum into your startup code (users may want to deploy quantum in 
something other than eventlet) and use the code fragment above to get 
the command line parameter.



>>         Thanks,
>>
>>         -Shiv
>>
>>
>>         _______________________________________________
>>         OpenStack-dev mailing list
>>         OpenStack-dev at lists.openstack.org
>>         <mailto:OpenStack-dev at lists.openstack.org>
>>         http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>
>>
>>
>
>
>
> _______________________________________________
> OpenStack-dev mailing list
> OpenStack-dev at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20121026/96679de3/attachment.html>