openstack/nova
Code Issues Proposed changes

Added a service heartbeat driver using Memcached.

Browse Source 
Today the heartbeat information of Nova services/nodes
is maintained in the DB, while each service updates the
corresponding record in the Service table periodically
(by default -- every 10 seconds), specifying the timestamp
of the last update. This mechanism is highly inefficient
and does not scale. E.g., maintaining the heartbeat
information for 1,000 nodes/services would require 100 DB
updates per second (just for the heartbeat).

This patch adds nova.servicegroup.drivers.memcached, a
service heartbeat driver using Memcached. You can reduce
DB updates with it.

blueprint memcached-service-heartbeat

Change-Id: I60bdb1cfbce1fea051f276ebfd6ccc4ad8fe6d2b
This commit is contained in:
Akira Yoshiyama
2013-02-09 14:26:04 +00:00
parent d980805880
commit 625e074033
4 changed files with 336 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
nova/servicegroup/api.py
View File

@@ -28,8 +28,10 @@ import random
LOG = logging.getLogger(__name__)
_default_driver = 'db'
servicegroup_driver_opt = cfg.StrOpt('servicegroup_driver',
default=_default_driver,
help='The driver for servicegroup service.')
default=_default_driver,
help='The driver for servicegroup '
'service (valid options are: '
'db, zk, mc)')
CONF = cfg.CONF
CONF.register_opt(servicegroup_driver_opt)
@@ -40,7 +42,8 @@ class API(object):
_driver = None
_driver_name_class_mapping = {
'db': 'nova.servicegroup.drivers.db.DbDriver',
'zk': 'nova.servicegroup.drivers.zk.ZooKeeperDriver'
'zk': 'nova.servicegroup.drivers.zk.ZooKeeperDriver',
'mc': 'nova.servicegroup.drivers.mc.MemcachedDriver'
}
def __new__(cls, *args, **kwargs):