Adds support for configurable ceph pool replication count

Fixes: bug 1251560

config.yaml

@@ -64,3 +64,13 @@ options:
         The name that will be used to create the Ceph's RBD image with. If the
         image name exists in Ceph, it will be re-used and the data will be
         overwritten.
+    ceph-osd-replication-count:
+        default: 2
+        type: int
+        description: |
+            This value dictates the number of replicas ceph must make of any
+            object it stores within the rabbitmq rbd pool. Of course, this only
+            applies if using Ceph as a backend store. Note that once the rabbitmq
+            rbd pool has been created, changing this value will not have any
+            effect (although it can be changed in ceph by manually configuring
+            your ceph cluster).

hooks/lib/ceph_utils.py

@@ -9,6 +9,7 @@
 #
 
 import commands
+import json
 import subprocess
 import os
 import shutil
@@ -66,16 +67,37 @@ def pool_exists(service, name):
     return name in out
 
 
-def create_pool(service, name):
-    cmd = [
-        'rados',
-        '--id',
-        service,
-        'mkpool',
-        name
-        ]
-    execute(cmd)
+def get_osds(service):
+    '''
+    Return a list of all Ceph Object Storage Daemons
+    currently in the cluster
+    '''
+    return json.loads(subprocess.check_output(['ceph', '--id', service,
+                                               'osd', 'ls', '--format=json']))
 
+def create_pool(service, name, replicas=2):
+    ''' Create a new RADOS pool '''
+    if pool_exists(service, name):
+        utils.juju_log('WARNING',
+                       "Ceph pool {} already exists, "
+                       "skipping creation".format(name))
+        return
+
+    # Calculate the number of placement groups based
+    # on upstream recommended best practices.
+    pgnum = (len(get_osds(service)) * 100 / replicas)
+    cmd = [
+        'ceph', '--id', service,
+        'osd', 'pool', 'create',
+        name, str(pgnum)
+    ]
+    subprocess.check_call(cmd)
+    cmd = [
+        'ceph', '--id', service,
+        'osd', 'pool', 'set', name,
+        'size', str(replicas)
+    ]
+    subprocess.check_call(cmd)
 
 def keyfile_path(service):
     return KEYFILE % service
@@ -220,7 +242,8 @@ def copy_files(src, dst, symlinks=False, ignore=None):
 
 
 def ensure_ceph_storage(service, pool, rbd_img, sizemb, mount_point,
-                        blk_device, fstype, system_services=[]):
+                        blk_device, fstype, system_services=[],
+                        rbd_pool_replicas=2):
     """
     To be called from the current cluster leader.
     Ensures given pool and RBD image exists, is mapped to a block device,
@@ -235,7 +258,7 @@ def ensure_ceph_storage(service, pool, rbd_img, sizemb, mount_point,
     # Ensure pool, RBD image, RBD mappings are in place.
     if not pool_exists(service, pool):
         utils.juju_log('INFO', 'ceph: Creating new pool %s.' % pool)
-        create_pool(service, pool)
+        create_pool(service, pool, replicas=rbd_pool_replicas)
 
     if not rbd_exists(service, pool, rbd_img):
         utils.juju_log('INFO', 'ceph: Creating RBD image (%s).' % rbd_img)

hooks/rabbitmq_server_relations.py

@@ -215,11 +215,13 @@ def ceph_changed():
         rbd_size = utils.config_get('rbd-size')
         sizemb = int(rbd_size.split('G')[0]) * 1024
         blk_device = '/dev/rbd/%s/%s' % (POOL_NAME, rbd_img)
+        rbd_pool_rep_count = utils.config_get('ceph-osd-replication-count')
         ceph.ensure_ceph_storage(service=SERVICE_NAME, pool=POOL_NAME,
                                  rbd_img=rbd_img, sizemb=sizemb,
                                  fstype='ext4', mount_point=RABBIT_DIR,
                                  blk_device=blk_device,
-                                 system_services=['rabbitmq-server'])
+                                 system_services=['rabbitmq-server'],
+                                 rbd_pool_replicas=rbd_pool_rep_count)
     else:
         utils.juju_log('INFO',
                        'This is not the peer leader. Not configuring RBD.')

revision

@@ -1 +1 @@
-97
+98