[train-squash] Remove hardcoded reference to cinder LVM loopback device

Backport note:
In CentOS 7, losetup does not include the -n option. We therefore use
a slightly different way to get the same output in this backport.

This is a squash commit of the following patches:

Rename loopback creates file

We are hitting bug[1] and it looks like because /dev/loop2 file
already exists the ansible task which creates loop didn't trigger.
Creating file with different name "/dev/loop2cinder" to pass the
issue[1]

[1] https://bugs.launchpad.net/tripleo/+bug/1872881

Change-Id: I4e18bee7864b71afca387fbea5857c9530faa2fc
Partial-Bug: #1872881
(cherry picked from commit f85caaf411)

Remove hardcoded reference to cinder LVM loopback device

Replace hardcoded reference to "/dev/loop2" by allowing losetup to
dynamically choose the next available loopback device number.

Eliminate hiera data that set cinder_lvm_loop_device_size. This is not
needed because the associated file is created by a host prep task, and
the puppet code will be removed by https://review.opendev.org/720139.

Closes-Bug: #1872881
Change-Id: Ia0302d1d55dcb8333d7db9713822075d32cd852a
(cherry picked from commit 502947b4be)

Improve the cinder LVM loopback device setup

The current implementation has the following issues:

1. It uses two tasks to get the loopback device information,
   and another to set a fact which is only used in the same
   play, which is unnecessary. This can all be done in a
   single task without the fact setting.

2. The setup of the loopback device is not idempotent. If any
   of the tasks fail, another device will be setup and the
   the tasks will continue to fail because it cannot handle
   multiple value returns.

3. The LVM PV & VG are setup using a shell task. This could
   be done using an Ansible module instead, which also makes
   the task idempotent.

4. The oneshot systemd unit starts very late in the boot cycle.
   It can be set to start much sooner, ensuring that the
   cinder-volume service starts successfully sooner.

5. The oneshot systemd unit starts a bash subshell unnecessarily.

This patch aims to resolve all these issues, improving the
execution time and idempotency of these tasks.

Closes-Bug: #1873899
Change-Id: I9421cf54f498b3f99a7e5afa11425a0b2419b399
(cherry picked from commit 580aca40e5)

[cinder-lvm] Resolve issue when there are multiple loop devices

If a deployment fails, there may end up being multiple loopback
devices pointing to the same image file. This currently breaks
the systemd unit file because it cannot handle starting two
devices properly.

We only need one device, so just in case this situation arises,
we now ensure that anything more than the first device is ignored
when running 'Get or create LVM loopback device'.

Change-Id: I012ee9c60bce4148da8ec6fe077ee20cfe34d555
Related-bug: #1873899
(cherry picked from commit d9152a601e)

deployment/cinder/cinder-common-container-puppet.yaml

@@ -106,36 +106,47 @@ outputs:
                     - 'M'
                 args:
                   creates: /var/lib/cinder/cinder-volumes
-              - name: cinder create LVM volume group
-                shell: |
-                  if ! losetup /dev/loop2; then
-                    losetup /dev/loop2 /var/lib/cinder/cinder-volumes
-                  fi
-                  if ! pvdisplay | grep cinder-volumes; then
-                    pvcreate /dev/loop2
-                  fi
-                  if ! vgdisplay | grep cinder-volumes; then
-                    vgcreate cinder-volumes /dev/loop2
+              - name: Get or create LVM loopback device
+                shell: |-
+                  exit_code=0
+                  existing_device=$(losetup -j /var/lib/cinder/cinder-volumes -l -O NAME | tail -n-1)
+                  if [[ -z "${existing_device}" ]]; then
+                      losetup -f /var/lib/cinder/cinder-volumes --show
+                      exit_code=2
+                  else
+                      echo ${existing_device%$'\n'*}
                   fi
+                  exit ${exit_code}
                 args:
                   executable: /bin/bash
-                  creates: /dev/loop2
+                register: _loopback_device
+                changed_when: _loopback_device.rc == 2
+                failed_when: _loopback_device.rc not in [0,2]
+              - name: Create LVM volume group
+                lvg:
+                  vg: "cinder-volumes"
+                  pvs: "{{ _loopback_device.stdout }}"
+                  state: present
               - name: cinder create service to run losetup for LVM on startup
                 copy:
                   dest: /etc/systemd/system/cinder-lvm-losetup.service
                   content: |
                     [Unit]
                     Description=Cinder LVM losetup
-                    After=syslog.target
+                    DefaultDependencies=no
+                    Conflicts=umount.target
+                    Requires=lvm2-lvmetad.service systemd-udev-settle.service
+                    Before=local-fs.target umount.target
+                    After=lvm2-lvmetad.service systemd-udev-settle.service
 
                     [Service]
                     Type=oneshot
-                    ExecStart=/bin/bash -c '/sbin/losetup /dev/loop2 || /sbin/losetup /dev/loop2 /var/lib/cinder/cinder-volumes'
-                    ExecStop=/bin/bash -c '/sbin/losetup -d /dev/loop2'
+                    ExecStart=/sbin/losetup {{ _loopback_device.stdout }} /var/lib/cinder/cinder-volumes
+                    ExecStop=/sbin/losetup -d {{ _loopback_device.stdout }}
                     RemainAfterExit=yes
 
                     [Install]
-                    WantedBy=multi-user.target
+                    WantedBy=local-fs-pre.target
               - name: cinder enable the LVM losetup service
                 systemd:
                   name: cinder-lvm-losetup

deployment/cinder/cinder-volume-container-puppet.yaml

@@ -92,10 +92,6 @@ parameters:
     default: iscsi
     description: Whether to use TCP ('iscsi') or iSER RDMA ('iser') for iSCSI
     type: string
-  CinderLVMLoopDeviceSize:
-    default: 10280
-    description: The size of the loopback file used by the cinder LVM driver.
-    type: number
   CinderNfsAvailabilityZone:
     default: ''
     description: >
@@ -212,7 +208,6 @@ outputs:
             tripleo::profile::base::cinder::volume::nfs::cinder_nfs_snapshot_support: {get_param: CinderNfsSnapshotSupport}
             tripleo::profile::base::cinder::volume::nfs::cinder_nas_secure_file_operations: {get_param: CinderNasSecureFileOperations}
             tripleo::profile::base::cinder::volume::nfs::cinder_nas_secure_file_permissions: {get_param: CinderNasSecureFilePermissions}
-            tripleo::profile::base::cinder::volume::iscsi::cinder_lvm_loop_device_size: {get_param: CinderLVMLoopDeviceSize}
             tripleo::profile::base::cinder::volume::iscsi::cinder_iscsi_helper: {get_param: CinderISCSIHelper}
             tripleo::profile::base::cinder::volume::iscsi::cinder_iscsi_protocol: {get_param: CinderISCSIProtocol}
             tripleo::profile::base::cinder::volume::rbd::cinder_rbd_ceph_conf: