New N2VC interface + updated libjuju
This commit introduces the Python3 N2VC module, which acts as a standard
interface to the VCA.
The goal of this is to provide a common way for modules to interface
with the VCA.
- Updated libjuju from 0.6.1 to 0.7.3
Signed-off-by: Adam Israel <adam.israel@canonical.com>
Change-Id: Ide70fb5ae5797eb6486de24653dc09a23f9c009e
diff --git a/modules/libjuju/tests/integration/test_model.py b/modules/libjuju/tests/integration/test_model.py
index 041f75a..ba2da92 100644
--- a/modules/libjuju/tests/integration/test_model.py
+++ b/modules/libjuju/tests/integration/test_model.py
@@ -1,11 +1,16 @@
import asyncio
+import mock
from concurrent.futures import ThreadPoolExecutor
from pathlib import Path
+
+from juju.client.client import ConfigValue, ApplicationFacade
+from juju.model import Model, ModelObserver
+from juju.utils import block_until, run_with_interrupt
+
import pytest
from .. import base
-from juju.model import Model
-from juju.client.client import ConfigValue
+
MB = 1
GB = 1024
@@ -18,16 +23,30 @@
from pathlib import Path
tests_dir = Path(__file__).absolute().parent.parent
bundle_path = tests_dir / 'bundle'
+ mini_bundle_file_path = bundle_path / 'mini-bundle.yaml'
async with base.CleanModel() as model:
await model.deploy(str(bundle_path))
+ await model.deploy(str(mini_bundle_file_path))
- for app in ('wordpress', 'mysql'):
+ for app in ('wordpress', 'mysql', 'myapp'):
assert app in model.applications
@base.bootstrapped
@pytest.mark.asyncio
+async def test_deploy_local_charm(event_loop):
+ from pathlib import Path
+ tests_dir = Path(__file__).absolute().parent.parent
+ charm_path = tests_dir / 'charm'
+
+ async with base.CleanModel() as model:
+ await model.deploy(str(charm_path))
+ assert 'charm' in model.applications
+
+
+@base.bootstrapped
+@pytest.mark.asyncio
async def test_deploy_bundle(event_loop):
async with base.CleanModel() as model:
await model.deploy('bundle/wiki-simple')
@@ -43,7 +62,7 @@
charm = 'cs:~johnsca/libjuju-test'
stable = await model.deploy(charm, 'a1')
edge = await model.deploy(charm, 'a2', channel='edge')
- rev = await model.deploy(charm+'-2', 'a3')
+ rev = await model.deploy(charm + '-2', 'a3')
assert [a.charm_url for a in (stable, edge, rev)] == [
'cs:~johnsca/libjuju-test-1',
@@ -111,17 +130,44 @@
# subordinates must be deployed without units
num_units=0,
)
- my_relation = await model.add_relation(
- 'ubuntu',
- 'nrpe',
- )
+
+ relation_added = asyncio.Event()
+ timeout = asyncio.Event()
+
+ class TestObserver(ModelObserver):
+ async def on_relation_add(self, delta, old, new, model):
+ if set(new.key.split()) == {'nrpe:general-info',
+ 'ubuntu:juju-info'}:
+ relation_added.set()
+ event_loop.call_later(2, timeout.set)
+
+ model.add_observer(TestObserver())
+
+ real_app_facade = ApplicationFacade.from_connection(model.connection())
+ mock_app_facade = mock.MagicMock()
+
+ async def mock_AddRelation(*args):
+ # force response delay from AddRelation to test race condition
+ # (see https://github.com/juju/python-libjuju/issues/191)
+ result = await real_app_facade.AddRelation(*args)
+ await relation_added.wait()
+ return result
+
+ mock_app_facade.AddRelation = mock_AddRelation
+
+ with mock.patch.object(ApplicationFacade, 'from_connection',
+ return_value=mock_app_facade):
+ my_relation = await run_with_interrupt(model.add_relation(
+ 'ubuntu',
+ 'nrpe',
+ ), timeout, event_loop)
assert isinstance(my_relation, Relation)
-async def _deploy_in_loop(new_loop, model_name):
- new_model = Model(new_loop)
- await new_model.connect_model(model_name)
+async def _deploy_in_loop(new_loop, model_name, jujudata):
+ new_model = Model(new_loop, jujudata=jujudata)
+ await new_model.connect(model_name)
try:
await new_model.deploy('cs:xenial/ubuntu')
assert 'ubuntu' in new_model.applications
@@ -138,7 +184,7 @@
with ThreadPoolExecutor(1) as executor:
f = executor.submit(
new_loop.run_until_complete,
- _deploy_in_loop(new_loop, model_name))
+ _deploy_in_loop(new_loop, model_name, model._connector.jujudata))
f.result()
await model._wait_for_new('application', 'ubuntu')
assert 'ubuntu' in model.applications
@@ -155,7 +201,7 @@
lambda: (
len(ghost.units) > 0 and
ghost.units[0].workload_status in terminal_statuses)
- )
+ )
# ghost will go in to blocked (or error, for older
# charm revs) if the resource is missing
assert ghost.units[0].workload_status == 'active'
@@ -174,7 +220,7 @@
lambda: (
len(ghost.units) > 0 and
ghost.units[0].workload_status in terminal_statuses)
- )
+ )
# ghost will go in to blocked (or error, for older
# charm revs) if the resource is missing
assert ghost.units[0].workload_status == 'active'
@@ -206,9 +252,8 @@
@pytest.mark.asyncio
async def test_watcher_reconnect(event_loop):
async with base.CleanModel() as model:
- await model.connection.ws.close()
- await asyncio.sleep(0.1)
- assert model.connection.is_open
+ await model.connection().ws.close()
+ await block_until(model.is_connected, timeout=3)
@base.bootstrapped