Hello
We have deployed Victoria release with kolla-ansible (openvswitch neutron driver), and after the reboot of control nodes we are experiencing issues with dhcp on some networks. I found out that there were missing dhcp tap interfaces in br-int openvswitch. After restart of dhcp agetn and openvswitch agent, interface was added to br-int and dhcp works again in that network. I found bug from 2016 (https://bugs.launchpad.net/kolla/+bug/1617188), that there is a script neutron-ovs-cleanup (it is in each neutron docker container), which should be run after reboot of dhcp/L3 nodes otherwise issues with dhcp/metadata/l3 functionalities could arise.
I do not see mentioned this in kolla documentation, bug is quite old and expired, so I would like to ask, is that still valid and should it be run manually after each reboot of dhcp/L3 node? If yes, what is best what how to achieve it when docker starts all container automatically after reboot. Should I stop neutron containers, run neutron-ovs-cleanup and start them again?
Kamil Madáč Slovensko IT a.s.