Recurring DNS lookup failures during devstack-gate-cleanup

[David Moreau Simard]

"Name or service not known -- Some packages may not be found!":

- http://logs.openstack.org/07/472607/145/check/tripleo-ci-centos-7-containers-multinode/ae132c1/logs/devstack-gate-cleanup-host.txt

- http://logs.openstack.org/77/524477/3/gate/tripleo-ci-centos-7-containers-multinode/54ca06d/logs/devstack-gate-cleanup-host.txt

- http://logs.openstack.org/51/524751/1/check-tripleo/tripleo-ci-centos-7-ovb-ha-oooq/fe3034f/logs/devstack-gate-cleanup-host.txt

- http://logs.openstack.org/83/524183/2/check/tripleo-ci-centos-7-undercloud-oooq/bf335fd/logs/devstack-gate-cleanup-host.txt

- http://logs.openstack.org/28/501028/41/check/tripleo-ci-centos-7-3nodes-multinode/8e744d3/logs/devstack-gate-cleanup-host.txt

- http://logs.openstack.org/45/523945/1/gate/tripleo-ci-centos-7-scenario007-multinode-oooq-container/c2d4c2f/logs/devstack-gate-cleanup-host.txt

- http://logs.openstack.org/40/523440/1/gate/tripleo-ci-centos-7-undercloud-oooq/9d53de6/logs/devstack-gate-cleanup-host.txt

[David Moreau Simard (dmsimard)]

Within the same job, DNS resolution will work and suddenly stop working.

Example:
#1
Works: http://logs.openstack.org/51/524751/1/check-tripleo/tripleo-ci-centos-7-ovb-ha-oooq/fe3034f/job-output.txt.gz#_2017-12-01_23_06_16_499864
Stops working: http://logs.openstack.org/51/524751/1/check-tripleo/tripleo-ci-centos-7-ovb-ha-oooq/fe3034f/job-output.txt.gz#_2017-12-01_23_08_05_983507

#2
Works: http://logs.openstack.org/77/524477/3/gate/tripleo-ci-centos-7-containers-multinode/54ca06d/job-output.txt.gz#_2017-12-02_21_46_22_319783
Stops working: http://logs.openstack.org/77/524477/3/gate/tripleo-ci-centos-7-containers-multinode/54ca06d/job-output.txt.gz#_2017-12-02_21_51_23_387262

#3
Works: http://logs.openstack.org/07/472607/145/check/tripleo-ci-centos-7-containers-multinode/ae132c1/job-output.txt.gz#_2017-12-02_23_17_43_106858
Stops working: http://logs.openstack.org/07/472607/145/check/tripleo-ci-centos-7-containers-multinode/ae132c1/job-output.txt.gz#_2017-12-02_23_21_38_186180

[David Moreau Simard (dmsimard)]

DNS resolution suspiciously stops working after a restart of unbound that occurs ~5 seconds before:
http://logs.openstack.org/07/472607/145/check/tripleo-ci-centos-7-containers-multinode/ae132c1/job-output.txt.gz#_2017-12-02_23_21_33_419832

[David Moreau Simard (dmsimard)]

Sagi sent a patch to remove the restart of Unbound: https://review.openstack.org/#/c/525228/