yahoo-eng-team team mailing list archive

[Sean Dague <sean@xxxxxxxxx>]

Public bug reported:

During some gate runs (especially multinode ones) we end up with the
following kinds of failures (all in this log
http://logs.openstack.org/66/258166/1/check/gate-tempest-dsvm-
full/41867ce/logs/syslog.txt.gz)

Dec 16 00:12:16 devstack-trusty-hpcloud-b3-6590764 dnsmasq-dhcp[25317]: not using configured address 10.1.0.4 because it is leased to fa:16:3e:8e:45:48
Dec 16 00:12:16 devstack-trusty-hpcloud-b3-6590764 dnsmasq-dhcp[25317]: DHCPDISCOVER(br100) fa:16:3e:8f:f1:18 no address available

fa:16:3e:8e:45:48 is a valid guest that we booted 45 seconds previously
(00:11:30). We destroy it at 00:11:35, but when doing so in nova-net we
get the following:

IP 10.1.0.4 leased that is not associated

http://logs.openstack.org/66/258166/1/check/gate-tempest-dsvm-
full/41867ce/logs/screen-n-net.txt.gz#_2015-12-16_00_11_35_298

Nova network knows something is wrong, dnsmasq has a lease that's no
good any more. This then causes a failure later when we try to boot
another node on that same address as dnsmasq rejects that.

** Affects: nova
     Importance: Undecided
         Status: New

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Compute (nova).
https://bugs.launchpad.net/bugs/1526737

Title:
  Race between nova-network and dnsmasq allocating ips

Status in OpenStack Compute (nova):
  New

Bug description:
  During some gate runs (especially multinode ones) we end up with the
  following kinds of failures (all in this log
  http://logs.openstack.org/66/258166/1/check/gate-tempest-dsvm-
  full/41867ce/logs/syslog.txt.gz)

  Dec 16 00:12:16 devstack-trusty-hpcloud-b3-6590764 dnsmasq-dhcp[25317]: not using configured address 10.1.0.4 because it is leased to fa:16:3e:8e:45:48
  Dec 16 00:12:16 devstack-trusty-hpcloud-b3-6590764 dnsmasq-dhcp[25317]: DHCPDISCOVER(br100) fa:16:3e:8f:f1:18 no address available

  fa:16:3e:8e:45:48 is a valid guest that we booted 45 seconds
  previously (00:11:30). We destroy it at 00:11:35, but when doing so in
  nova-net we get the following:

  IP 10.1.0.4 leased that is not associated

  http://logs.openstack.org/66/258166/1/check/gate-tempest-dsvm-
  full/41867ce/logs/screen-n-net.txt.gz#_2015-12-16_00_11_35_298

  Nova network knows something is wrong, dnsmasq has a lease that's no
  good any more. This then causes a failure later when we try to boot
  another node on that same address as dnsmasq rejects that.

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1526737/+subscriptions