[openstack-dev] [Neutron][Requirement] SQLAlchemy raises None, causes TypeError

2013-07-18 Thread Nachi Ueno
Hi folks Sometimes, I got SQLAlchemy raises None, causes TypeError This get debugging hard, because it hides real reasons. It looks like the problem of combining eventlet and sqlalchemy. http://stackoverflow.com/questions/9525220/sqlalchemy-raises-none-causes-typeerror Latest version of

Re: [openstack-dev] [Neutron][Requirement] SQLAlchemy raises None, causes TypeError

2013-07-18 Thread Eugene Nikanorov
Hi Nachi, Usually that exception is raised when something goes wrong during transaction commit (__exit__ of with block that we usually use) I do session.flush() manually to see specific exception in such cases. In case exception is potentially unavoidable, explicitly calling flush() seems to be a

Re: [openstack-dev] [Neutron][Requirement] SQLAlchemy raises None, causes TypeError

2013-07-18 Thread Nachi Ueno
Hi Eugene Thanks for tips. I'm trying to investigate this one https://bugs.launchpad.net/neutron/+bug/1202797 Since we are using with context.session.. without flush for many place, it is good to fix this lib side bug. I have tested devstack with sqlalcemy 0.7.10 but it looks like keystone

Re: [openstack-dev] [Neutron][Requirement] SQLAlchemy raises None, causes TypeError

2013-07-18 Thread David Ripton
On 07/18/2013 03:48 PM, Nachi Ueno wrote: Sometimes, I got SQLAlchemy raises None, causes TypeError This get debugging hard, because it hides real reasons. It looks like the problem of combining eventlet and sqlalchemy.

Re: [openstack-dev] [Neutron][Requirement] SQLAlchemy raises None, causes TypeError

2013-07-18 Thread Nachi Ueno
Hi David Thanks for your information. hmm it looks long way.. 2013/7/18 David Ripton drip...@redhat.com: On 07/18/2013 03:48 PM, Nachi Ueno wrote: Sometimes, I got SQLAlchemy raises None, causes TypeError This get debugging hard, because it hides real reasons. It looks like the problem of