Comment 4 for bug 1979089

@minsel no that's not quite right. What I have is a production env where there are circa 280 HA routers at any one time. Sometimes routers get deleted (at least once a day lets say) and due to this bug, if during the process of the router being deleted, the vrrp transitions the master, you may hit this issue. The "thousands a day" bit refers to the amount of times the l3-agent tries to unsuccessfully spawn a new haproxy for the deleted router i.e. it keeps doing it over and over and over because when it tries to spawn it fails since the namespace is no longer there and then the metadata monitor notices it failed and tries again (and it never gives up). Hope that makes sense.