@Adam: Bug #1006898 seems unlikely IMHO. We are not running with 'multi_host' nor with VLANS. So there is one dnsmasq process.
The following is just conjecture as the event has rotated out of the logs and I have not done any further testing. These events happen rather infrequently, but are a significant annoyance to be noticed. The dnsmasq(8) man page implies that dnsmasq will block while running the dhcp-script. Ordinarily this would not be a problem for individual leases, however if the daemon receives a HUP signal it will parse the current lease database and run the dhcp-script with the 'old' action for each lease. Assuming the above is correct and we have a dnsmasq lease file of 200 records and a runtime of 0.15s per invocation¹ of dhcp-script, then the daemon could be blocking for up to 30 seconds (or longer). If a host was renewing an expired lease at the time, it is conceivable that the lease may expire. The default lease length is 120s which seems to correlate with the logs provided above. I am not sure if any of this helps. [1]: Non-significant mean run time determined by running 'time /usr/bin/nova-dhcpbridge' ten times. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1026621 Title: nova-network gets release_fixed_ip events from someplace, but the database still keeps them associated with instances To manage notifications about this bug go to: https://bugs.launchpad.net/nova/+bug/1026621/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
