Retrieve keyrings for openstack clients from ceph cluster fails
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack-Ansible |
Invalid
|
Undecided
|
Unassigned |
Bug Description
I am running a new deploy of Rocky 18.1.5 on Ubuntu 18.04 hosts, following the ceph example configuration options.
openstack-ansible setup-hosts.yml completes successfully
openstack-ansible setup-infrastru
openstack-ansible setup-openstack.yml (specifically os-glance-
"Ceph_Client: Retrieve keyrings for openstack clients from ceph cluster"
when it fails with the following:
failed: [infra1_
"attempts": 3,
"changed": false,
"cmd": "ceph auth get client.glance >/dev/null && ceph auth get-or-create client.glance",
"delta": "0:00:00.004645",
"end": "2019-03-27 18:42:48.097148",
"invocation": {
"warn": true
}
},
"item": "glance",
"msg": "non-zero return code",
"rc": 127,
"start": "2019-03-27 18:42:48.092503",
"stderr": "/bin/sh: 1: ceph: not found",
"stderr_lines": [
"/bin/sh: 1: ceph: not found"
],
"stdout": "",
"stdout_lines": []
user_variables.yml contains:
ceph_pkg_source: ceph
ceph_stable_
## Common Ceph Overrides
ceph_mons:
- 172.29.236.11
- 172.29.236.12
- 172.29.236.13
## Ceph cluster fsid (must be generated before first run)
## Generate a uuid using: python -c 'import uuid; print(str(
generate_fsid: false
fsid: 31371b52-
## ceph-ansible settings
## See https:/
## additional configuration options availble.
monitor_
public_network: "172.29.236.0/22"
cluster_network: "172.29.244.0/22"
osd_scenario: collocated
osd_auto_discovery: true
journal_size: 10240 # size in MB
# ceph-ansible automatically creates pools & keys for OpenStack services
openstack_config: true
cinder_ceph_client: cinder
#cinder_
glance_ceph_client: glance
glance_
glance_
glance_
nova_libvirt_
This will also fail if I do not include the ceph.conf in the user_variables.yml:
"[ceph_client : Get ceph.conf and store contents when ceph_conf_file is not defined]"
fatal: [infra1_
I resolved this by removing the "ceph_mons" override
It wouldn't hurt to add to the documentation that this is only needed for external ceph clusters.