Deploying a Ceph Nautilus cluster in a rdo standalone scenario, an issue appear during the monitor bootstrap phase.
In particular, using the standard network config proposed and supported, the mon is unable to start election and form the quorum because it's stuck with the following messages:
7fb411007700 0 -- [v2:192.168.24.1:3300/0,v1:192.168.24.1:6789/0] send_to message mon_probe(probe 4b5c8c0a-ff60-454b-a1b4-9747aa737d19 name s
tandalone new mon_release 14) v7 with empty dest
debug 2019-04-09 07:47:25.826 7fb411007700 0 -- [v2:192.168.24.1:3300/0,v1:192.168.24.1:6789/0] send_to message mon_probe(probe 4b5c8c0a-ff60-454b-a1b4-9747aa737d19 name standalone new mon_release 14) v7 with empty dest
debug 2019-04-09 07:47:27.826 7fb411007700 0 -- [v2:192.168.24.1:3300/0,v1:192.168.24.1:6789/0] send_to message mon_probe(probe 4b5c8c0a-ff60-454b-a1b4-9747aa737d19 name standalone new mon_release 14) v7 with empty dest
Running the same jobs but applying the hack on br-ex described here [1]:
https://review.openstack.org/#/c/651231/2/deployment/ceph-ansible/ceph-base.yaml
the cluster is able to perform the election and the deploy ends correctly.
Here an example of red ci [2] during a standalone execution
[2] RED CI:
https://logs.rdoproject.org/21/18721/29/check/rdoinfo-tripleo-stein-centos-7-scenario001-standalone/28c108e/logs/undercloud/home/zuul/undercloud-ansible-tjWQlx/ceph-ansible/ceph_ansible_command.log.txt.gz
WORKAROUND
parameter_defaults: traConfig:
CephAnsibleEx
mon_host_v1: { 'enabled': False }