Comment 9 for bug 1333814

The reason the fix from https://review.openstack.org/102381 is insufficient is that the old problem with ceph.conf reordering is back in ceph-deploy 1.5.x.

In https://github.com/ceph/ceph-deploy/blob/v1.5.2/ceph_deploy/hosts/remotes.py#L68 write_conf() does a string comparison of the expected configuration with verbatim contents of target ceph.conf, where expected string matches the actual ceph.conf but is reordered.

Expected:

[global]
auth_service_required = cephx
osd_pool_default_size = 1
filestore_xattr_use_omap = true
osd_pool_default_min_size = 1
auth_client_required = cephx
osd_pool_default_pg_num = 256
auth_cluster_required = cephx
mon_host = 192.168.0.2
osd_journal_size = 2048
public_network = 192.168.0.0/24
mon_initial_members = node-1
osd_mkfs_type = xfs
auth_supported = cephx
cluster_network = 192.168.1.0/24
fsid = ea1533ec-4eaa-498d-a6b3-b10236bfd21b

Actual:

[global]
auth_service_required = cephx
filestore_xattr_use_omap = true
auth_client_required = cephx
auth_cluster_required = cephx
mon_host = 192.168.0.2
mon_initial_members = node-1
fsid = ea1533ec-4eaa-498d-a6b3-b10236bfd21b
osd_pool_default_size = 1
osd_pool_default_min_size = 1
osd_pool_default_pg_num = 256
public_network = 192.168.0.0/24
osd_journal_size = 2048
auth_supported = cephx
osd_mkfs_type = xfs
cluster_network = 192.168.1.0/24

This only happens on CentOS 6.x (might be due to Python 2.6), on Ubuntu "ceph-deploy mon create" completes successfully.