[openstack-dev] move deallocate port from after hypervisor driver does detach interface when doing detach_interface
taget at linux.vnet.ibm.com
Thu Oct 23 02:26:16 UTC 2014
? 2014?10?22? 23:18, Ben Nemec ??:
> Please don't send review requests to the list. The preferred methods of
> asking for reviews are discussed in this post:
thanks for letting me know this. and I've noticed it.
but I am not sending review request, I'd like to bring up an discussion
what is the correct behavior when we get the
currently , user can not notice he detach an interface failed, the
exception was not handled,
and the port is deleted in neutron but guest still can see it.
I add the patch link just show you my propose, for better to understand,
I'd like more people to discuss this.
> On 10/22/2014 02:57 AM, Eli Qiao wrote:
>> hi all.
>> when I am reviewing code of in nova/compute/manager.py
>> I find that the detach_interface calls deallocate port from neutron first
>> then calls detach_interface in the hypervisor, then what will happen if
>> hypervisor detach_interface failed? the result is the port can be seen
>> on guest but removed from neutron, this seems inconsistent.
>> I submit a patch  to propose remove port from neutron side after
>> detach_interface successfully, and keep neutron port if catch exception when
>> detach_interface then give a log message.
>> can some one kindly help to take a look at it and give some comments?
>>  https://review.openstack.org/#/c/130151/
>> OpenStack-dev mailing list
>> OpenStack-dev at lists.openstack.org
> OpenStack-dev mailing list
> OpenStack-dev at lists.openstack.org
Eli (Li Yong) Qiao
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the OpenStack-dev