openstack/manila
Code Issues Proposed changes

Implement Ensure Shares in the CephFS driver

Browse Source 
This allows switching between NFS protocol helpers
that the driver currently supports and avoids
unnecessary ensuring of driver resources on restarts
unless a small set of config opts are toggled.

Use of either NFS protocol helper in the driver
will trigger access rules to be reapplied when
ensuring shares.

Implements: bp use-cephadm-nfs-ganesha
Partial-Bug: #2035137
Change-Id: I93c4e8616a1cfb5ab713b420aff69464969f28d5
Signed-off-by: Goutham Pacha Ravi <gouthampravi@gmail.com>
This commit is contained in:
Goutham Pacha Ravi 2023-03-29 22:19:19 -07:00
parent 389a2ea1e5
commit ee8af90109
4 changed files with 126 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

65
manila/share/drivers/cephfs/driver.py
View File

@ -138,6 +138,13 @@ cephfs_opts = [
cfg.StrOpt('cephfs_filesystem_name',
help="The name of the filesystem to use, if there are "
"multiple filesystems in the cluster."),
cfg.StrOpt('cephfs_ensure_all_shares_salt',
default="manila_cephfs_reef_bobcat",
help="Provide a unique string value to make the driver "
"ensure all of the shares it has created during "
"startup. Ensuring would re-export shares and this "
"action isn't always required, unless something has "
"been administratively modified on CephFS.")
]
cephfsnfs_opts = [
@ -540,15 +547,31 @@ class CephFSDriver(driver.ExecuteMixin, driver.GaneshaMixin,
context, share, access_rules, add_rules, delete_rules,
share_server=share_server)
def ensure_share(self, context, share, share_server=None):
def get_backend_info(self, context):
return self.protocol_helper.get_backend_info(context)
def ensure_shares(self, context, shares):
share_updates = {}
for share in shares:
share_updates[share['id']] = {
'reapply_access_rules':
self.protocol_helper.reapply_rules_while_ensuring_shares,
}
try:
export_location = self._get_export_locations(share)
share_updates[share['id']].update({
'export_locations': self._get_export_locations(share),
})
except exception.ShareBackendException as e:
if 'does not exist' in str(e).lower():
raise exception.ShareResourceNotFound(share_id=share['id'])
raise
return export_location
msg = ("Share instance %(si)s belonging to share "
"%(share)s cannot be found on the backend.")
msg_payload = {'si': share['id'],
'share': share['share_id']}
LOG.exception(msg, msg_payload)
share_updates[share['id']] = {
'status': constants.STATUS_ERROR,
}
return share_updates
def extend_share(self, share, new_size, share_server=None):
# resize FS subvolume/share
@ -778,6 +801,7 @@ class NativeProtocolHelper(ganesha.NASHelperBase):
supported_access_types = (CEPHX_ACCESS_TYPE, )
supported_access_levels = (constants.ACCESS_LEVEL_RW,
constants.ACCESS_LEVEL_RO)
reapply_rules_while_ensuring_shares = False
def __init__(self, execute, config, **kwargs):
self.rados_client = kwargs.pop('rados_client')
@ -803,6 +827,12 @@ class NativeProtocolHelper(ganesha.NASHelperBase):
return result
def get_backend_info(self, context):
return {
"cephfs_ensure_all_shares_salt":
self.configuration.cephfs_ensure_all_shares_salt,
}
def get_export_locations(self, share, subvolume_path):
# To mount this you need to know the mon IPs and the path to the volume
mon_addrs = self.get_mon_addrs()
@ -1055,6 +1085,7 @@ class NFSProtocolHelper(NFSProtocolHelperMixin, ganesha.GaneshaNASHelper2):
shared_data = {}
supported_protocols = ('NFS',)
reapply_rules_while_ensuring_shares = True
def __init__(self, execute, config_object, **kwargs):
if config_object.cephfs_ganesha_server_is_remote:
@ -1151,12 +1182,22 @@ class NFSProtocolHelper(NFSProtocolHelperMixin, ganesha.GaneshaNASHelper2):
return export_ips
def get_backend_info(self, context):
backend_info = {
"cephfs_ganesha_export_ips": self.config.cephfs_ganesha_export_ips,
"cephfs_ganesha_server_ip": self.config.cephfs_ganesha_server_ip,
"cephfs_ensure_all_shares_salt":
self.configuration.cephfs_ensure_all_shares_salt,
}
return backend_info
class NFSClusterProtocolHelper(NFSProtocolHelperMixin, ganesha.NASHelperBase):
supported_access_types = ('ip', )
supported_access_levels = (constants.ACCESS_LEVEL_RW,
constants.ACCESS_LEVEL_RO)
reapply_rules_while_ensuring_shares = True
def __init__(self, execute, config_object, **kwargs):
self.rados_client = kwargs.pop('rados_client')
@ -1308,3 +1349,15 @@ class NFSClusterProtocolHelper(NFSProtocolHelperMixin, ganesha.NASHelperBase):
self._deny_access(share)
return rule_state_map
def get_backend_info(self, context):
backend_info = {
"cephfs_ganesha_export_ips":
self.configuration.cephfs_ganesha_export_ips,
"cephfs_ganesha_server_ip":
self.configuration.cephfs_ganesha_server_ip,
"cephfs_nfs_cluster_id": self.nfs_clusterid,
"cephfs_ensure_all_shares_salt":
self.configuration.cephfs_ensure_all_shares_salt,
}
return backend_info

6
manila/share/drivers/ganesha/__init__.py
View File

@ -53,6 +53,12 @@ class NASHelperBase(metaclass=abc.ABCMeta):
delete_rules, share_server=None):
"""Update access rules of share."""
def get_backend_info(self, context):
raise NotImplementedError
def ensure_shares(self, context, shares):
raise NotImplementedError
class GaneshaNASHelper(NASHelperBase):
"""Perform share access changes using Ganesha version < 2.4."""

63
manila/tests/share/drivers/cephfs/test_driver.py
View File

@ -235,38 +235,53 @@ class CephFSDriverTestCase(test.TestCase):
self._context, self._share, access_rules, add_rules, delete_rules,
share_server=None)
def test_ensure_share(self):
expected_exports = {
def test_ensure_shares(self):
self._driver.protocol_helper.reapply_rules_while_ensuring_shares = True
shares = [
fake_share.fake_share(share_id='123', share_proto='NFS'),
fake_share.fake_share(share_id='456', share_proto='NFS'),
fake_share.fake_share(share_id='789', share_proto='NFS')
]
export_locations = [
{
'path': '1.2.3.4,5.6.7.8:/foo/bar',
'is_admin_only': False,
'metadata': {},
},
{
'path': '1.2.3.4,5.6.7.8:/foo/quz',
'is_admin_only': False,
'metadata': {},
},
]
expected_updates = {
shares[0]['id']: {
'status': constants.STATUS_ERROR,
'reapply_access_rules': True,
},
shares[1]['id']: {
'export_locations': export_locations[0],
'reapply_access_rules': True,
},
shares[2]['id']: {
'export_locations': export_locations[1],
'reapply_access_rules': True,
}
}
err_message = (f"Error ENOENT: subvolume {self._share['id']} does "
f"not exist")
expected_exception = exception.ShareBackendException(err_message)
self.mock_object(
self._driver, '_get_export_locations',
mock.Mock(return_value=expected_exports))
mock.Mock(side_effect=[expected_exception] + export_locations))
returned_exports = self._driver.ensure_share(self._context,
self._share)
actual_updates = self._driver.ensure_shares(self._context, shares)
self._driver._get_export_locations.assert_called_once_with(self._share)
self.assertEqual(expected_exports, returned_exports)
def test_ensure_share_subvolume_not_found(self):
message = f"Error ENOENT: subvolume {self._share['id']} does not exist"
expected_exception = exception.ShareBackendException(message)
self.mock_object(
self._driver, '_get_export_locations',
mock.Mock(side_effect=expected_exception))
self.assertRaises(
exception.ShareResourceNotFound,
self._driver.ensure_share,
self._context,
self._share)
self._driver._get_export_locations.assert_called_once_with(self._share)
self.assertEqual(3, self._driver._get_export_locations.call_count)
self._driver._get_export_locations.assert_has_calls([
mock.call(shares[0]), mock.call(shares[1]), mock.call(shares[2])])
self.assertEqual(expected_updates, actual_updates)
def test_delete_share(self):
clone_status_prefix = "fs clone status"

17
releasenotes/notes/bug-2035137-cephfs-support-ensure-shares-b72fe18381af274a.yaml Normal file
View File

@ -0,0 +1,17 @@
---
fixes:
- |
The CephFS backend driver now supports a bulk share recovery mechanism
(``ensure_shares``). At startup time, a combination of driver configuration
options will determine if the driver must re-evaluate export paths of
previously created shares. If these configuration options do not change,
service startup will skip through this recovery stage.
- |
The CephFS backend driver will also reapply access rules when performing
a recovery of pre-existing shares.
upgrade:
- |
A new configuration option called ``cephfs_ensure_all_shares_salt`` has
been introduced to assist cloud administrtrators that would like the CephFS
driver to reconcile export paths of existing shares during service
startup.