Fixes Ceph PG calculation

Baremetal deployments were failing because the ceph PG size was
exceeding the max allowed. Virtual was still working because we lower
the number of pools and pg/osd. This patch changes the values to a
number which should work for both virtual and baremetal. Also includes a
fix which adds the controllers back as OSDs and a few other cleanup
issues.

JIRA: APEX-614
JIRA: APEX-569

Change-Id: I2ad65727ecdcaa0454eb53d25e32b7f1a53cd3a4
Signed-off-by: Tim Rozet <trozet@redhat.com>

diff --git a/apex/overcloud/deploy.py b/apex/overcloud/deploy.py
index 8367ffa..a45b3a9 100644 (file)
--- a/apex/overcloud/deploy.py
+++ b/apex/overcloud/deploy.py
@@ -423,7 +423,7 @@ def prep_image(ds, ns, img, tmp_dir, root_pw=None, docker_tag=None,
                                            docker_tag=docker_tag))
     # if containers with ceph, and no ceph device we need to use a
     # persistent loop device for Ceph OSDs
-    if docker_tag and not ds_opts.get('ceph_device', None):
+    if docker_tag and ds_opts['ceph_device'] == '/dev/loop3':
         tmp_losetup = os.path.join(tmp_dir, 'losetup.service')
         with open(tmp_losetup, 'w') as fh:
             fh.write(LOSETUP_SERVICE)
@@ -685,28 +685,26 @@ def prep_storage_env(ds, ns, virtual, tmp_dir):
         ceph_params = {
             'DockerCephDaemonImage': docker_image,
         }
-        if not ds['global_params']['ha_enabled']:
-            ceph_params['CephPoolDefaultSize'] = 1
 
+        # max pgs allowed are calculated as num_mons * 200. Therefore we
+        # set number of pgs and pools so that the total will be less:
+        # num_pgs * num_pools * num_osds
+        ceph_params['CephPoolDefaultSize'] = 2
+        ceph_params['CephPoolDefaultPgNum'] = 32
         if virtual:
             ceph_params['CephAnsibleExtraConfig'] = {
                 'centos_package_dependencies': [],
                 'ceph_osd_docker_memory_limit': '1g',
                 'ceph_mds_docker_memory_limit': '1g',
             }
-            ceph_params['CephPoolDefaultPgNum'] = 32
-        if 'ceph_device' in ds_opts and ds_opts['ceph_device']:
-            ceph_device = ds_opts['ceph_device']
-        else:
-            # TODO(trozet): make this DS default after Fraser
-            ceph_device = '/dev/loop3'
-
+        ceph_device = ds_opts['ceph_device']
         ceph_params['CephAnsibleDisksConfig'] = {
             'devices': [ceph_device],
             'journal_size': 512,
             'osd_scenario': 'collocated'
         }
         utils.edit_tht_env(storage_file, 'parameter_defaults', ceph_params)
+    # TODO(trozet): remove following block as we only support containers now
     elif 'ceph_device' in ds_opts and ds_opts['ceph_device']:
         with open(storage_file, 'a') as fh:
             fh.write('  ExtraConfig:\n')

diff --git a/apex/settings/deploy_settings.py b/apex/settings/deploy_settings.py
index 4f887ed..29fe64f 100644 (file)
--- a/apex/settings/deploy_settings.py
+++ b/apex/settings/deploy_settings.py
@@ -26,11 +26,11 @@ REQ_DEPLOY_SETTINGS = ['sdn_controller',
                        'os_version',
                        'l2gw',
                        'sriov',
-                       'containers']
+                       'containers',
+                       'ceph_device']
 
 OPT_DEPLOY_SETTINGS = ['performance',
                        'vsperf',
-                       'ceph_device',
                        'yardstick',
                        'dovetail',
                        'odl_vpp_routing_node',
@@ -105,6 +105,8 @@ class DeploySettings(dict):
                     self['deploy_options'][req_set] = 'ovs'
                 elif req_set == 'ceph':
                     self['deploy_options'][req_set] = True
+                elif req_set == 'ceph_device':
+                    self['deploy_options'][req_set] = '/dev/loop3'
                 elif req_set == 'odl_version':
                     self['deploy_options'][req_set] = \
                         constants.DEFAULT_ODL_VERSION

diff --git a/apex/tests/test_apex_overcloud_deploy.py b/apex/tests/test_apex_overcloud_deploy.py
index b69c44d..f1db91a 100644 (file)
--- a/apex/tests/test_apex_overcloud_deploy.py
+++ b/apex/tests/test_apex_overcloud_deploy.py
@@ -499,7 +499,7 @@ class TestOvercloudDeploy(unittest.TestCase):
             'DockerCephDaemonImage':
                 '192.0.2.1:8787/ceph/daemon:tag-build-master-luminous-centos'
                 '-7',
-            'CephPoolDefaultSize': 1,
+            'CephPoolDefaultSize': 2,
             'CephAnsibleExtraConfig': {
                 'centos_package_dependencies': [],
                 'ceph_osd_docker_memory_limit': '1g',

diff --git a/lib/ansible/playbooks/deploy_overcloud.yml b/lib/ansible/playbooks/deploy_overcloud.yml
index 90c237b..53dc699 100644 (file)
--- a/lib/ansible/playbooks/deploy_overcloud.yml
+++ b/lib/ansible/playbooks/deploy_overcloud.yml
@@ -57,6 +57,14 @@
         owner: root
         group: root
       become: yes
+    - name: Insert Ceph OSDs into Controller role
+      lineinfile:
+        path: /usr/share/openstack-tripleo-heat-templates/roles_data.yaml
+        insertbefore:  '^\s*-\sOS::TripleO::Services::CephRbdMirror$'
+        line: '    - OS::TripleO::Services::CephOSD'
+        owner: root
+        group: root
+      become: yes
     - name: Upload glance images
       shell: "{{ stackrc }} && openstack overcloud image upload"
       become: yes