[k8s][HA][dns] DNS pod should run more than 1 on different nodes

[Feilong Wang]

DNS is a very critical service in k8s world, so it would be nice to have it replicate more than 1 and on differents nodes. A lot of discussion about this:

https://github.com/kubernetes/kubeadm/issues/128
https://github.com/kubernetes/kubernetes/issues/40063
https://github.com/kubernetes/kops/issues/2693

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/555223

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/555223
Committed: https://git.openstack.org/cgit/openstack/magnum/commit/?id=54a4ac9f8453716d84150d4fe0b88e65dba24fbf
Submitter: Zuul
Branch: master

commit 54a4ac9f8453716d84150d4fe0b88e65dba24fbf
Author: Feilong Wang <email address hidden>
Date: Thu Mar 22 22:38:56 2018 +1300

    Make DNS pod autoscale

    DNS service is a very critical service in k8s world, though it's not
    a part of k8s itself. So it would be nice to have it replicate more
    than 1 and on differents nodes to have high availbility. Otherwise,
    services running on k8s cluster will be broken if the node contains
    DNS pod down. Another sample is, when user would like to do a cluster
    upgrade, services will be borken when the node containers DNS pod
    being replaced. You can find lots of discussion about this, please
    refer [1],[2] and [3].

    [1] https://github.com/kubernetes/kubeadm/issues/128
    [2] https://github.com/kubernetes/kubernetes/issues/40063
    [3] https://github.com/kubernetes/kops/issues/2693

    Closes-Bug: #1757554

    Change-Id: Ic64569d4bdcf367955398d5badef70e7afe33bbb

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/queens
Review: https://review.openstack.org/565345

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/565345
Committed: https://git.openstack.org/cgit/openstack/magnum/commit/?id=2fc72e9b0f189c5c105fabddd8fca66c8212b8d0
Submitter: Zuul
Branch: stable/queens

commit 2fc72e9b0f189c5c105fabddd8fca66c8212b8d0
Author: Feilong Wang <email address hidden>
Date: Thu Mar 22 22:38:56 2018 +1300

    Make DNS pod autoscale

    DNS service is a very critical service in k8s world, though it's not
    a part of k8s itself. So it would be nice to have it replicate more
    than 1 and on differents nodes to have high availbility. Otherwise,
    services running on k8s cluster will be broken if the node contains
    DNS pod down. Another sample is, when user would like to do a cluster
    upgrade, services will be borken when the node containers DNS pod
    being replaced. You can find lots of discussion about this, please
    refer [1],[2] and [3].

    [1] https://github.com/kubernetes/kubeadm/issues/128
    [2] https://github.com/kubernetes/kubernetes/issues/40063
    [3] https://github.com/kubernetes/kops/issues/2693

    Closes-Bug: #1757554

    Change-Id: Ic64569d4bdcf367955398d5badef70e7afe33bbb
    (cherry picked from commit 54a4ac9f8453716d84150d4fe0b88e65dba24fbf)

[Spyros Trigazis (strigazi)]

Cluster creations are broken after this, the manifest of coredns became too large for user_data. Let's convert this to a software deployment.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/566533

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/magnum 6.1.1 release.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/566533
Committed: https://git.openstack.org/cgit/openstack/magnum/commit/?id=38878fd642c7d04c65a229ce55da1c4d68efbb19
Submitter: Zuul
Branch: master

commit 38878fd642c7d04c65a229ce55da1c4d68efbb19
Author: Spyros Trigazis <email address hidden>
Date: Mon May 7 09:41:46 2018 +0000

    k8s_fedora: Make CoreDNS config a SoftwareDeployment

    After adding the autoscaler for coredns, the limit
    for user_data was reached again. Make coredns
    config a SoftwareDeployment.

    Change-Id: I0a9852e9293842e859947acf0c4b6da20394436a
    Closes-Bug: #1757554

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/magnum 7.0.0 release.