if /run/keepalived.pid file exists, keepalive wont start

[Steven Dake]

See summary. If the containers fail for some reason or cleanup-containers -fv is fun, keepalive will no longer start. One easy solution is to remove the pid file for keepalived. We don't really need it. ATM we specify it as a config option.

[Steven Dake (sdake)]

This wouldn't be such a problem but the location where the pid file is stored is bindmounted to /run:/run.

[Steven Dake (sdake)]

My preferred solution to this problem is to remove /run:/run as bindmounts, and store /run as a data container shared between haproxy and keepalived.

[Steven Dake (sdake)]

note even sharing /run in comment #2, pid file still needs removal.

[Steven Dake (sdake)]

Apparently this only happens after an unclean shutdown of the container, (such as a crash or container-cleanup -fv). Unclear if we can remove the --pid flag to keepalived and still have a working system in the cases of HA. Recommend testing if the file exists, and if it does, delete it at container start time.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/215380

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/215380
Committed: https://git.openstack.org/cgit/stackforge/kolla/commit/?id=129f1d06a141c75643336399181ea157fb0dd82a
Submitter: Jenkins
Branch: master

commit 129f1d06a141c75643336399181ea157fb0dd82a
Author: Duane DeCapite <email address hidden>
Date: Thu Aug 20 19:51:57 2015 -0400

    Workaround for existing keepalived.pid file

    Fixes issue where keepalived will not restart after container failure.
    Removes keepalived.pid file if one exists on container startup.

    Change-Id: I6ec3297a5e49a4d5255323aa7f75dfdc10d0c5ec
    Closes-Bug: #1485079