cloud-init resets eth0 of supplemental node breaking networking

[Harry Rybacki]

Description
===========

cloud-init is resetting the config in /etc/system/networking-scripts/ for eth0 -- probably on the first boot after the guest is provisioned. Default behavior is to assign the 10th host on the `external_network_cidr` as a static IP on this guest. After cloud-init resets the config, the adapter is picking up a new IP via DHCP

  # Created by cloud-init on instance boot automatically, do not edit.
  #
  BOOTPROTO=dhcp
  DEVICE=eth0
  HWADDR=52:54:00:ed:20:3b
  ONBOOT=yes
  TYPE=Ethernet
  USERCTL=no

thereby invalidating what is in a given deployment's ssh.config.ansible file

  <snip>
  Host supplemental
    ProxyCommand ssh -o UserKnownHostsFile=/dev/null -o StrictHostKeyChecking=no -o
      ConnectTimeout=60 -i /home/alee/.quickstart/id_rsa_virt_power <email address hidden> -W 192.168.23.10:22
    IdentityFile /home/XXX/.quickstart/id_rsa_supplemental
    User stack
    StrictHostKeyChecking no
    UserKnownHostsFile=/dev/null
  <snip>

and stopping playbook execution when it next attempts to reach the guest

  TASK [freeipa-setup : Create FreeIPA deployment script] ************************
  task path: /home/XXX/.quickstart/usr/local/share/ansible/roles/freeipa-setup/tasks/main.yml:14
  Wednesday 20 September 2017 20:23:39 -0400 (0:00:00.088) 0:50:41.051 ***
  <supplemental> ssh_retry: attempt: 0, ssh return code is 255. cmd (/bin/sh -c 'echo ~stack &&
  sleep 0'...), pausing for 0 seconds
  <supplemental> ssh_retry: attempt: 1, ssh return code is 255. cmd (/bin/sh -c 'echo ~stack &&
  sleep 0'...), pausing for 1 seconds
  <supplemental> ssh_retry: attempt: 2, ssh return code is 255. cmd (/bin/sh -c 'echo ~stack &&
  sleep 0'...), pausing for 3 seconds
  fatal: [supplemental]: UNREACHABLE! => {"changed": false, "msg": "Failed to connect to the host
  via ssh: Warning: Permanently added 'XXX.com,10.13.57.27' (ECDSA) to the list of known
  hosts.\r\nchannel 0: open failed: connect failed: No route to host\r\nstdio forwarding
  failed\r\nssh_exchange_identification: Connection closed by remote host\r\n", "unreachable":
  true}

Steps to reproduce
==================

Deploy quickstart w/ IPA against a CentOS virthost

# Deploy w/ quickstart
pushd tripleo-quickstart
bash quickstart.sh \
  --bootstrap \
  --ansible-debug \
  --no-clone \
  --playbook quickstart-extras.yml \
  --working-dir $WORKING_DIR \
  --release master-tripleo-ci \
  --config $WORKSPACE/config/general_config/ipa.yml \
  --nodes $WORKSPACE/config/nodes/1ctlr_2comp_1supp.yml \
  --teardown "all" \
  --tags "all" \
  $VIRTHOST

Expected result
===============

Quickstart runs successfully and the FreeIPA server exists on the supplemental node.

Actual result
=============

Quickstart fails upon attempting to reach the supplemental node after it has started.

Environment
===========

release=master
CentOS 7.3
cloud-init-0.7.9-9.el7.centos.2.x86_64

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/508158

[Harry Rybacki (hrybacki-h)]

Confirmed that a vanilla quickstart run does deploy successfully. So something in my invocation is resulting in the issue.

[Harry Rybacki (hrybacki-h)]

Bug resolved in https://review.openstack.org/#/c/510007/

[OpenStack Infra (hudson-openstack)]

Change abandoned by Harry Rybacki (<email address hidden>) on branch: master
Review: https://review.openstack.org/508158
Reason: Abandoned in lieu of: https://review.openstack.org/#/c/510007/