Merge lp:~xfactor973/charm-helpers/ceph-pool-replicas into lp:charm-helpers

[Chris Holcombe]

The change adds in a call to update the ceph pool once the pool creation is finished. By not having this call I broke the cinder and glance charm. The pools were getting created with the wrong number of replicas and never allowed those placement groups in the pool to go healthy in Oil testing.

[Jason Hobbs (jason-hobbs)]

+1 - I tested this change and it fixed the problem for me. I was able to successfully upload an image to glance.

review: Approve

[Chris MacNaughton (chris.macnaughton)]

+11 - Looks good to me

review: Approve

[Ryan Beisner (1chb1n)]

Thanks for the fix, and thanks for verifying in OIL. Will merge, then storage charmers will need to resync and propose changes ASAP so we can revalidate in 16.04 pre-release tests.

review: Approve

=== modified file 'charmhelpers/contrib/storage/linux/ceph.py'
--- charmhelpers/contrib/storage/linux/ceph.py	2016-03-09 18:23:35 +0000
+++ charmhelpers/contrib/storage/linux/ceph.py	2016-04-13 21:49:16 +0000
@@ -221,6 +221,10 @@
                    self.name, str(self.pg_num)]
             try:
                 check_call(cmd)
+                # Set the pool replica size
+                update_pool(client=self.service,
+                            pool=self.name,
+                            settings={'size': str(self.replicas)})
             except CalledProcessError:
                 raise
 

=== modified file 'tests/contrib/storage/test_linux_ceph.py'
--- tests/contrib/storage/test_linux_ceph.py	2016-03-09 18:23:35 +0000
+++ tests/contrib/storage/test_linux_ceph.py	2016-04-13 21:49:16 +0000
@@ -234,6 +234,7 @@
 
         self.check_call.assert_has_calls([
             call(['ceph', '--id', 'admin', 'osd', 'pool', 'create', 'test', str(200)]),
+            call(['ceph', '--id', 'admin', 'osd', 'pool', 'set', 'test', 'size', str(3)])
         ])
 
     @patch.object(ceph_utils, 'get_osds')