openstack/octavia
Code Issues Proposed changes

Create disabled members in haproxy

Browse Source 
Members that were disabled / admin_state_up=False were simply excluded
from the haproxy configuration we pass to the amps. Instead, we should
be creating them in a disabled state, so they return in health messages
as status "maint", and can be marked OFFLINE via the standard health
mechanism, instead of just via override hacks.
This also resolves a bug introduced in an earlier change:
https://review.openstack.org/#/c/567322/
which caused admin-downed members to stay in NO_MONITOR always.

Change-Id: I6615b3ff89d7cef2af52d474aab3a03d947f98be
This commit is contained in:
Adam Harwell 2018-05-11 15:11:20 -07:00
parent db69c051ac
commit c6ff8596db
5 changed files with 108 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
octavia/common/jinja/haproxy/templates/macros.j2
View File

@ -184,9 +184,15 @@ frontend {{ listener.id }}
{% else %}
{% set member_backup_opt = "" %}
{% endif %}
{{ "server %s %s:%d weight %s%s%s%s%s"|e|format(
{% if member.enabled %}
{% set member_enabled_opt = "" %}
{% else %}
{% set member_enabled_opt = " disabled" %}
{% endif %}
{{ "server %s %s:%d weight %s%s%s%s%s%s"|e|format(
member.id, member.address, member.protocol_port, member.weight,
hm_opt, persistence_opt, proxy_protocol_opt, member_backup_opt)|trim() }}
hm_opt, persistence_opt, proxy_protocol_opt, member_backup_opt,
member_enabled_opt)|trim() }}
{% endmacro %}
@ -257,7 +263,7 @@ backend {{ pool.id }}
option allbackups
timeout connect {{ listener.timeout_member_connect }}
timeout server {{ listener.timeout_member_data }}
{% for member in pool.members if member.enabled %}
{% for member in pool.members %}
{{- member_macro(constants, pool, member) -}}
{% endfor %}
{% endmacro %}

20
octavia/tests/unit/common/jinja/haproxy/test_jinja_cfg.py
View File

@ -339,6 +339,26 @@ class TestHaproxyCfg(base.TestCase):
self.assertEqual(sample_configs.sample_base_expected_config(
backend=be), rendered_obj)
def test_render_template_disabled_member(self):
be = ("backend sample_pool_id_1\n"
" mode http\n"
" balance roundrobin\n"
" cookie SRV insert indirect nocache\n"
" fullconn 98\n"
" option allbackups\n"
" timeout connect 5000\n"
" timeout server 50000\n"
" server sample_member_id_1 10.0.0.99:82 weight 13 "
"cookie sample_member_id_1\n"
" server sample_member_id_2 10.0.0.98:82 weight 13 "
"cookie sample_member_id_2 disabled\n\n")
rendered_obj = self.jinja_cfg.render_loadbalancer_obj(
sample_configs.sample_amphora_tuple(),
sample_configs.sample_listener_tuple(proto='HTTP', monitor=False,
disabled_member=True))
self.assertEqual(sample_configs.sample_base_expected_config(
backend=be), rendered_obj)
def test_render_template_ping_monitor_http(self):
be = ("backend sample_pool_id_1\n"
" mode http\n"

9
octavia/tests/unit/common/sample_configs/sample_configs.py
View File

@ -420,6 +420,7 @@ def sample_listener_tuple(proto=None, monitor=True, persistence=True,
l7=False, enabled=True, insert_headers=None,
be_proto=None, monitor_ip_port=False,
monitor_proto=None, backup_member=False,
disabled_member=False,
timeout_client_data=50000,
timeout_member_connect=5000,
timeout_member_data=50000,
@ -465,7 +466,7 @@ def sample_listener_tuple(proto=None, monitor=True, persistence=True,
persistence_type=persistence_type,
persistence_cookie=persistence_cookie,
monitor_ip_port=monitor_ip_port, monitor_proto=monitor_proto,
backup_member=backup_member)]
backup_member=backup_member, disabled_member=disabled_member)]
l7policies = []
return in_listener(
id='sample_listener_id_1',
@ -541,7 +542,8 @@ def sample_tls_container_tuple(id='cont_id_1', certificate=None,
def sample_pool_tuple(proto=None, monitor=True, persistence=True,
persistence_type=None, persistence_cookie=None,
sample_pool=1, monitor_ip_port=False,
monitor_proto=None, backup_member=False):
monitor_proto=None, backup_member=False,
disabled_member=False):
proto = 'HTTP' if proto is None else proto
monitor_proto = proto if monitor_proto is None else monitor_proto
in_pool = collections.namedtuple(
@ -557,7 +559,8 @@ def sample_pool_tuple(proto=None, monitor=True, persistence=True,
monitor_ip_port=monitor_ip_port),
sample_member_tuple('sample_member_id_2', '10.0.0.98',
monitor_ip_port=monitor_ip_port,
backup=backup_member)]
backup=backup_member,
enabled=not disabled_member)]
if monitor is True:
mon = sample_health_monitor_tuple(proto=monitor_proto)
elif sample_pool == 2:

75
octavia/tests/unit/controller/healthmanager/health_drivers/test_update_db.py
View File

@ -96,6 +96,11 @@ class TestUpdateHealthDb(base.TestCase):
mock_pool1 = mock.Mock()
mock_pool1.id = "pool-id-1"
mock_pool1.members = []
if health_monitor:
mock_hm1 = mock.Mock()
mock_pool1.health_monitor = mock_hm1
else:
mock_pool1.health_monitor = None
mock_lb.pools = [mock_pool1]
if mock_listener1:
mock_listener1.pools = [mock_pool1]
@ -125,7 +130,7 @@ class TestUpdateHealthDb(base.TestCase):
"recv_time": time.time()
}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
@ -243,7 +248,7 @@ class TestUpdateHealthDb(base.TestCase):
"recv_time": time.time()
}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
self.hm.update_health(health)
@ -289,7 +294,7 @@ class TestUpdateHealthDb(base.TestCase):
"recv_time": time.time()
}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
@ -324,7 +329,7 @@ class TestUpdateHealthDb(base.TestCase):
"recv_time": time.time()
}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
mock_member2 = mock.Mock()
@ -416,7 +421,7 @@ class TestUpdateHealthDb(base.TestCase):
"recv_time": time.time()
}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
self.hm.update_health(health)
@ -447,7 +452,7 @@ class TestUpdateHealthDb(base.TestCase):
"members": {"member-id-1": constants.DRAIN}}}}},
"recv_time": time.time()}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
self.hm.update_health(health)
@ -487,7 +492,7 @@ class TestUpdateHealthDb(base.TestCase):
"members": {"member-id-1": constants.MAINT}}}}},
"recv_time": time.time()}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
self.hm.update_health(health)
@ -527,7 +532,7 @@ class TestUpdateHealthDb(base.TestCase):
"members": {"member-id-1": "blah"}}}}},
"recv_time": time.time()}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
self.hm.update_health(health)
@ -563,7 +568,7 @@ class TestUpdateHealthDb(base.TestCase):
"recv_time": time.time()
}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
@ -606,7 +611,7 @@ class TestUpdateHealthDb(base.TestCase):
"recv_time": time.time()
}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree(health_monitor=False))
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
@ -629,6 +634,48 @@ class TestUpdateHealthDb(base.TestCase):
self.member_repo.update.assert_not_called()
def test_update_health_member_down_no_hm(self):
health = {
"id": self.FAKE_UUID_1,
"listeners": {
"listener-id-1": {"status": constants.OPEN, "pools": {
"pool-id-1": {"status": constants.UP,
"members": {"member-id-1": constants.MAINT}
}
}
}
},
"recv_time": time.time()
}
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree(health_monitor=False))
mock_members[0].admin_state_up = False
mock_members[0].operating_status = constants.NO_MONITOR
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
self.hm.update_health(health)
self.assertTrue(self.amphora_health_repo.replace.called)
# test listener, member
for listener_id, listener in six.iteritems(
health.get('listeners', {})):
self.listener_repo.update.assert_any_call(
self.session_mock, listener_id,
operating_status=constants.ONLINE)
for pool_id, pool in six.iteritems(listener.get('pools', {})):
self.hm.pool_repo.update.assert_any_call(
self.session_mock, pool_id,
operating_status=constants.ONLINE)
self.member_repo.update.assert_any_call(
self.session_mock, mock_members[0].id,
operating_status=constants.OFFLINE)
def test_update_health_member_no_check(self):
health = {
@ -645,7 +692,7 @@ class TestUpdateHealthDb(base.TestCase):
"recv_time": time.time()
}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
self.hm.update_health(health)
@ -731,7 +778,7 @@ class TestUpdateHealthDb(base.TestCase):
"recv_time": time.time()
}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
self.hm.update_health(health)
@ -782,7 +829,7 @@ class TestUpdateHealthDb(base.TestCase):
"recv_time": time.time()
}
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]
@ -924,7 +971,7 @@ class TestUpdateHealthDb(base.TestCase):
self.hm.loadbalancer_repo.update.side_effect = (
[sqlalchemy.orm.exc.NoResultFound])
mock_lb, mock_listener1, mock_pool1, mock_member1 = (
mock_lb, mock_listener1, mock_pool1, mock_members = (
self._make_mock_lb_tree())
self.amphora_repo.get_all_lbs_on_amphora.return_value = [mock_lb]

12
releasenotes/notes/render-disabled-members-to-make-statuses-consistent-69189f71da2e02e8.yaml Normal file
View File

@ -0,0 +1,12 @@
---
fixes:
- |
Creating a member on a pool with no healthmonitor would sometimes briefly
update their operating status from `NO_MONITOR` to `OFFLINE` and back to
`NO_MONITOR` during the provisioning sequence. This flapping will no longer
occur.
- |
Members that are disabled via `admin_state_up=False` are now rendered in
the HAProxy configuration on the amphora as `disabled`. Previously they
were not rendered at all. This means that disabled members will now
appear in health messages, and will properly change status to OFFLINE.