some of the disks allocated for Ceph OSD are not provisioned

[Dmitry Borodaenko]

The node in question has multiple hard drives fully allocated to Ceph OSD via Fuel CLI using JSON like follows:

{
                "name": "sdc",
                "free_space": 2860428,
                "volumes": [
                    {
                        "vg": "os",
                        "type": "pv",
                        "lvm_meta_size": 0,
                        "size": 0
                    },
                    {
                        "partition_guid": "45b0969e-9b03-4f30-b4c6-b4b80ceff106",
                        "name": "cephjournal",
                        "mount": "none",
                        "disk_label": "cephjournal",
                        "type": "partition",
                        "file_system": "ext4",
                        "size": 0
                    },
                    {
                        "partition_guid": "4fbd7e29-9d25-41b8-afd0-062c0ceff05d",
                        "name": "ceph",
                        "mount": "none",
                        "disk_label": "cephosd",
                        "type": "partition",
                        "file_system": "xfs",
                        "size": 2860428
                    }
                ],
                "type": "disk",
                "id": "disk/by-path/pci-0000:02:00.0-scsi-0:2:2:0",
                "size": 2861056
            },

Some of these disks are formatted as XFS during provisioning and then registered with Ceph during deployment, for others mkfs.xfs fails:

chroot /mnt/sysimage sgdisk --typecode=3:4fbd7e29-9d25-41b8-afd0-062c0ceff05d /dev/disk/by-path/pci-0000:02:00.0-scsi-0:2:4:0
+ chroot /mnt/sysimage sgdisk --typecode=3:4fbd7e29-9d25-41b8-afd0-062c0ceff05d /dev/disk/by-path/pci-0000:02:00.0-scsi-0:2:4:0
The operation has completed successfully.
mkfs.xfs -f /dev/sde3 -L cephosd
+ mkfs.xfs -f /dev/sde3 -L cephosd
meta-data=/dev/sde3 isize=256 agcount=4, agsize=183067392 blks
         = sectsz=512 attr=2, projid32bit=0
data = bsize=4096 blocks=732269568, imaxpct=5
         = sunit=0 swidth=0 blks
naming =version 2 bsize=4096 ascii-ci=0
log =internal log bsize=4096 blocks=357553, version=2
         = sectsz=512 sunit=0 blks, lazy-count=1
realtime =none extsz=4096 blocks=0, rtextents=0
echo 'UUID=cd9f9932-8d29-4da7-aa49-ee296c9a799a none xfs defaults 0 0' >> /mnt/sysimage/etc/fstab
+ echo 'UUID=cd9f9932-8d29-4da7-aa49-ee296c9a799a none xfs defaults 0 0'
chroot /mnt/sysimage sgdisk --typecode=3:4fbd7e29-9d25-41b8-afd0-062c0ceff05d /dev/disk/by-path/pci-0000:02:00.0-scsi-0:2:5:0
+ chroot /mnt/sysimage sgdisk --typecode=3:4fbd7e29-9d25-41b8-afd0-062c0ceff05d /dev/disk/by-path/pci-0000:02:00.0-scsi-0:2:5:0
The operation has completed successfully.
mkfs.xfs -f /dev/sdf3 -L cephosd
+ mkfs.xfs -f /dev/sdf3 -L cephosd
/dev/sdf3: No such file or directory

Since the disks for which mkfs failed do not get a 'cephosd' partition label, they are not passed to the ceph::osd Puppet manifest and are not registered with Ceph.

[Ryan Moe (rmoe)]

https://review.openstack.org/#/c/54841/
https://github.com/Mirantis/fuel/pull/773