diff --git a/neutron/agent/l3/ha.py b/neutron/agent/l3/ha.py
index 2caa9994911..01d408e869e 100644
--- a/neutron/agent/l3/ha.py
+++ b/neutron/agent/l3/ha.py
@@ -45,6 +45,13 @@ OPTS = [
     cfg.IntOpt('ha_vrrp_advert_int',
                default=2,
                help=_('The advertisement interval in seconds')),
+    cfg.IntOpt('ha_keepalived_state_change_server_threads',
+               default=(1 + common_utils.cpu_count()) // 2,
+               min=1,
+               help=_('Number of concurrent threads for '
+                      'keepalived server connection requests.'
+                      'More threads create a higher CPU load '
+                      'on the agent node.')),
 ]
 
 
@@ -79,7 +86,8 @@ class L3AgentKeepalivedStateChangeServer(object):
 
     def run(self):
         server = agent_utils.UnixDomainWSGIServer(
-            'neutron-keepalived-state-change')
+            'neutron-keepalived-state-change',
+            num_threads=self.conf.ha_keepalived_state_change_server_threads)
         server.start(KeepalivedStateChangeHandler(self.agent),
                      self.get_keepalived_state_change_socket_path(self.conf),
                      workers=0,
diff --git a/neutron/agent/linux/utils.py b/neutron/agent/linux/utils.py
index 93524b92a89..d66a73966f9 100644
--- a/neutron/agent/linux/utils.py
+++ b/neutron/agent/linux/utils.py
@@ -402,11 +402,12 @@ class UnixDomainHttpProtocol(eventlet.wsgi.HttpProtocol):
 
 
 class UnixDomainWSGIServer(wsgi.Server):
-    def __init__(self, name):
+    def __init__(self, name, num_threads=None):
         self._socket = None
         self._launcher = None
         self._server = None
-        super(UnixDomainWSGIServer, self).__init__(name, disable_ssl=True)
+        super(UnixDomainWSGIServer, self).__init__(name, disable_ssl=True,
+                                                   num_threads=num_threads)
 
     def start(self, application, file_socket, workers, backlog, mode=None):
         self._socket = eventlet.listen(file_socket,
diff --git a/neutron/tests/unit/agent/linux/test_utils.py b/neutron/tests/unit/agent/linux/test_utils.py
index 15c4af9f675..b6dbde64bcf 100644
--- a/neutron/tests/unit/agent/linux/test_utils.py
+++ b/neutron/tests/unit/agent/linux/test_utils.py
@@ -411,9 +411,9 @@ class TestUnixDomainWSGIServer(base.BaseTestCase):
         super(TestUnixDomainWSGIServer, self).setUp()
         self.eventlet_p = mock.patch.object(utils, 'eventlet')
         self.eventlet = self.eventlet_p.start()
-        self.server = utils.UnixDomainWSGIServer('test')
 
     def test_start(self):
+        self.server = utils.UnixDomainWSGIServer('test')
         mock_app = mock.Mock()
         with mock.patch.object(self.server, '_launch') as launcher:
             self.server.start(mock_app, '/the/path', workers=5, backlog=128)
@@ -427,6 +427,7 @@ class TestUnixDomainWSGIServer(base.BaseTestCase):
             launcher.assert_called_once_with(mock_app, workers=5)
 
     def test_run(self):
+        self.server = utils.UnixDomainWSGIServer('test')
         self.server._run('app', 'sock')
 
         self.eventlet.wsgi.server.assert_called_once_with(
@@ -436,3 +437,17 @@ class TestUnixDomainWSGIServer(base.BaseTestCase):
             log=mock.ANY,
             max_size=self.server.num_threads
         )
+
+    def test_num_threads(self):
+        num_threads = 8
+        self.server = utils.UnixDomainWSGIServer('test',
+                                                 num_threads=num_threads)
+        self.server._run('app', 'sock')
+
+        self.eventlet.wsgi.server.assert_called_once_with(
+            'sock',
+            'app',
+            protocol=utils.UnixDomainHttpProtocol,
+            log=mock.ANY,
+            max_size=num_threads
+        )
diff --git a/releasenotes/notes/keepalived-state-change-server-threads-9ed775e7533dd1a0.yaml b/releasenotes/notes/keepalived-state-change-server-threads-9ed775e7533dd1a0.yaml
new file mode 100644
index 00000000000..b2d3bd1eaa7
--- /dev/null
+++ b/releasenotes/notes/keepalived-state-change-server-threads-9ed775e7533dd1a0.yaml
@@ -0,0 +1,10 @@
+---
+upgrade:
+  - A new option ``ha_keepalived_state_change_server_threads`` has been
+    added to configure the number of concurrent threads spawned for
+    keepalived server connection requests. Higher values increase the
+    CPU load on the agent nodes. The default value is half of the number
+    of CPUs present on the node. This allows operators to tune the
+    number of threads to suit their environment. With more threads,
+    simultaneous requests for multiple HA routers state change can be
+    handled faster.