Description of problem: During osp13 update we got one of our customer that had a failure during the update run on the controllers.  The octavia container were not starting correctly and failed with that kind of error:


INFO:__main__:Setting permission for /var/log/octavia/octavia.log
++ cat /run_command
+ CMD='/usr/bin/octavia-worker --config-file /usr/share/octavia/octavia-dist.conf --config-file /etc/octavia/octavia.conf --log-file /var/log/octavia/worker.log --config-file /etc/octavia/post-deploy.conf --config-dir /etc/octavia/conf.d/octavia-worker'
+ ARGS=
+ [[ ! -n '' ]]
+ . kolla_extend_start
++ set -o errexit
++ OCTAVIA_LOG_DIR=/var/log/kolla/octavia
++ [[ ! -d /var/log/kolla/octavia ]]
+++ stat -c %a /var/log/kolla/octavia
++ [[ 2755 != \7\5\5 ]]
++ chmod 755 /var/log/kolla/octavia
++ . /usr/local/bin/kolla_octavia_extend_start
+ echo 'Running command: '\''/usr/bin/octavia-worker --config-file /usr/share/octavia/octavia-dist.conf --config-file /etc/octavia/octavia.conf --log-file /var/log/octavia/worker.log --config-file /etc/octavia/post-deploy.conf --config-dir /etc/octavia/conf.d/octavia-worker'\'''
Running command: '/usr/bin/octavia-worker --config-file /usr/share/octavia/octavia-dist.conf --config-file /etc/octavia/octavia.conf --log-file /var/log/octavia/worker.log --config-file /etc/octavia/post-deploy.conf --config-dir /etc/octavia/conf.d/octavia-worker'
+ exec /usr/bin/octavia-worker --config-file /usr/share/octavia/octavia-dist.conf --config-file /etc/octavia/octavia.conf --log-file /var/log/octavia/worker.log --config-file /etc/octavia/post-deploy.conf --config-dir /etc/octavia/conf.d/octavia-worker
Traceback (most recent call last):
  File "/usr/bin/octavia-worker", line 10, in <module>
    sys.exit(main())
  File "/usr/lib/python2.7/site-packages/octavia/cmd/octavia_worker.py", line 30, in main
    octavia_service.prepare_service(sys.argv)
  File "/usr/lib/python2.7/site-packages/octavia/common/service.py", line 25, in prepare_service
    config.init(argv[1:])
  File "/usr/lib/python2.7/site-packages/octavia/common/config.py", line 631, in init
    **kwargs)
  File "/usr/lib/python2.7/site-packages/oslo_config/cfg.py", line 2504, in __call__
    raise ConfigFilesNotFoundError(self._namespace._files_not_found)
oslo_config.cfg.ConfigFilesNotFoundError: Failed to find some config files: /etc/octavia/post-deploy.conf


Then when converge was applied (this is the deploy command re-applied) the container configuration was fix and started.

This is certainly related to https://review.opendev.org/#/c/691936 and releted https://review.opendev.org/#/c/691935/ .  In that patch we change the default configuration path.  But those tasks are only done during deployment.  So the command fails during update and is settle during converge.  Those roles “octavia-*” seems to be handled by mistral directly.  Not sure exactly during the converge (deploy) process this is triggered.

We need to be able to make those adjustment during update, either with a specific update tasks, or in the puppet run (not sure it's used to create the new file though).

With a update tasks we run it on the host, the it won't be easy to create, a puppet configuration would be much easier.