I guess we should provide some workaround for 6.0 and document it in release notes. If one of controllers in HA environment is destroyed/deleted all OS services work slowly, especially Horizon dashboard, due to delays from keystone. Such issue can be manually fixed in the following way:
1. determine the IP address assigned to management interface of unavailable controller node (should be in /etc/hosts on alive nodes);
2. remove this address from list of memcached servers in keystone.conf on each controller (`DOWN_IP="192.168.0.5"; sed -r 's/'$DOWN_IP':11211,?//g' -i.bak /etc/keystone/keystone.conf`)
3. restart keystone daemon on all controllers (`service openstack-keystone restart || restart keystone`)
I guess we should provide some workaround for 6.0 and document it in release notes. If one of controllers in HA environment is destroyed/deleted all OS services work slowly, especially Horizon dashboard, due to delays from keystone. Such issue can be manually fixed in the following way:
1. determine the IP address assigned to management interface of unavailable controller node (should be in /etc/hosts on alive nodes); IP="192. 168.0.5" ; sed -r 's/'$DOWN_ IP':11211, ?//g' -i.bak /etc/keystone/ keystone. conf`)
2. remove this address from list of memcached servers in keystone.conf on each controller (`DOWN_
3. restart keystone daemon on all controllers (`service openstack-keystone restart || restart keystone`)