[openstack-dev] [neutron] replace external ip monitor

2015-05-04 Thread Peter V. Saveliev
… Hello. I would like to discuss the possibility to replace external ip monitor in the neutron code [1] with an internal native Python code [2] The issues of the current implementation: * an external process management * text output parsing (possibly buffered) The proposed library: * pure

Re: [openstack-dev] [neutron] replace external ip monitor

2015-05-04 Thread Assaf Muller
. - Original Message - … Hello. I would like to discuss the possibility to replace external ip monitor in the neutron code [1] with an internal native Python code [2] The issues of the current implementation: * an external process management * text output parsing (possibly

Re: [openstack-dev] [neutron] replace external ip monitor

2015-05-04 Thread Miguel Ángel Ajo
Does the library require root privileges to work for the operations you’re planning to do? That would be a stopper, since all the agents run unprivileged, and all the operations are filtered by the oslo root wrap daemon or cmdline tool. Best, Miguel Ángel. Miguel Ángel Ajo On Monday, 4 de

Re: [openstack-dev] [neutron] replace external ip monitor

2015-05-04 Thread Peter V. Saveliev
On 05/04/2015 03:25 PM, Miguel Ángel Ajo wrote: Does the library require root privileges to work for the operations you’re planning to do? Nope. Only the network stack changes need CAP_NET_ADMIN (add/del address, interface, route, traffic queue etc), and netns operations need the root

Re: [openstack-dev] [neutron] replace external ip monitor

2015-05-04 Thread Peter V. Saveliev
On 05/04/2015 02:58 PM, Assaf Muller wrote: skip / How's packaging looking on all supported platforms? It is packaged already for Fedora, RHEL, Debian, Ubuntu, Gentoo as soon as I know. Maybe for some other platforms. Pypi is provided as well. On a related note, ip_monitor.py is 87

Re: [openstack-dev] [neutron] replace external ip monitor

2015-05-04 Thread Peter V. Saveliev
On 05/04/2015 02:58 PM, Assaf Muller wrote: skip / errata, should be read as: … also has more than 200 functional tests in the regression testing cycle. pardonnez-moi, multiple editions of the mail sometimes lead to mistakes. skip/ -- Peter V. Saveliev