def test__host_draining_resumes_on_hostmgr_recovery(host_affinity_job,
maintenance, resmgr,
hostmgr):
# Pick a host that is UP and start maintenance on it
test_host = get_host_in_state(hpb.HOST_STATE_UP)
# Set host affinity of the job to the selected host
host_affinity_job.job_config.defaultConfig.\
constraint.labelConstraint.label.value = test_host
host_affinity_job.create()
host_affinity_job.wait_for_state(goal_state='RUNNING')
def all_running():
return all(t.state == task.RUNNING
for t in host_affinity_job.get_tasks().values())
host_affinity_job.wait_for_condition(all_running)
constraint = host_affinity_job.job_config.defaultConfig.constraint
test_host = constraint.labelConstraint.label.value
# Stop resmgr to ensure maintenance queue is not polled
resmgr.stop()
resp = maintenance['start']([test_host])
assert resp
hostmgr.restart()
resmgr.start()
# Wait for host to transition to DOWN
host.wait_for_host_state(test_host, hpb.HOST_STATE_DOWN)
def test__host_maintenance_lifecycle(host_affinity_job, maintenance):
# Pick a host that is UP and start maintenance on it
test_host = get_host_in_state(hpb.HOST_STATE_UP)
# Set host affinity of the job to the selected host
host_affinity_job.job_config.defaultConfig.\
constraint.labelConstraint.label.value = test_host
host_affinity_job.create()
# Start maintenance on the selected host
resp = maintenance['start']([test_host])
assert resp
assert host.is_host_in_state(test_host, hpb.HOST_STATE_DRAINING)
# Wait for host to transition to DOWN
host.wait_for_host_state(test_host, hpb.HOST_STATE_DOWN)
# Complete maintenance on the test hosts
resp = maintenance['stop']([test_host])
assert resp
# Host should no longer be DOWN
assert not host.is_host_in_state(test_host, hpb.HOST_STATE_DOWN)
host.wait_for_host_state(test_host, hpb.HOST_STATE_UP)
def clean_up():
# kill stateless jobs. This is needed since host draining
# is done in SLA aware manner for stateless jobs.
for j in stateless_query_jobs(client=client[0]):
j.stop()
if not draining_hosts:
return
for h in draining_hosts:
wait_for_host_state(h, host_pb2.HOST_STATE_DOWN)
stop(draining_hosts)
def clean_up():
if not draining_hosts:
return
for h in draining_hosts:
wait_for_host_state(h, host_pb2.HOST_STATE_DOWN)
stop(draining_hosts)