def test_workflow_spec_cache_update_via_workbook_service(self):
wb_text = """
version: '2.0'
name: wb
workflows:
wf:
tasks:
task1:
action: std.echo output="Echo"
"""
wb_service.create_workbook_v2(wb_text)
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(0, spec_parser.get_wf_definition_spec_cache_size())
wf = db_api.get_workflow_definition('wb.wf')
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wf.id, wf.updated_at)
self.assertEqual(1, len(wf_spec.get_tasks()))
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
# Now update workflow definition and check that cache is updated too.
wb_text = """
version: '2.0'
name: wb
workflows:
wf:
tasks:
task1:
action: std.echo output="1"
task2:
action: std.echo output="2"
"""
wb_service.update_workbook_v2(wb_text)
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
wf = db_api.get_workflow_definition(wf.id)
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wf.id, wf.updated_at)
self.assertEqual(2, len(wf_spec.get_tasks()))
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(2, spec_parser.get_wf_definition_spec_cache_size())
def test_cache_workflow_spec_no_duplicates(self):
wfs_text = """
version: '2.0'
wf:
tasks:
task1:
action: std.noop
on-success:
- task2
- task3
task2:
workflow: sub_wf my_param="val1"
task3:
workflow: sub_wf my_param="val2"
sub_wf:
input:
- my_param
tasks:
task1:
action: std.echo output="Param value is <% $.my_param %>"
"""
wfs = wf_service.create_workflows(wfs_text)
self.assertEqual(2, len(wfs))
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(0, spec_parser.get_wf_definition_spec_cache_size())
wf_ex = self.engine.start_workflow('wf')
self.await_workflow_success(wf_ex.id)
# We expect to have a cache entry for every workflow execution
# but two of them should refer to the same object.
self.assertEqual(3, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(2, spec_parser.get_wf_definition_spec_cache_size())
sub_wf_execs = db_api.get_workflow_executions(name='sub_wf')
self.assertEqual(2, len(sub_wf_execs))
spec1 = spec_parser.get_workflow_spec_by_execution_id(
sub_wf_execs[0].id
)
spec2 = spec_parser.get_workflow_spec_by_execution_id(
sub_wf_execs[1].id
)
self.assertIs(spec1, spec2)
def test_cache_workflow_spec_no_duplicates(self):
wfs_text = """
version: '2.0'
wf:
tasks:
task1:
action: std.noop
on-success:
- task2
- task3
task2:
workflow: sub_wf my_param="val1"
task3:
workflow: sub_wf my_param="val2"
sub_wf:
input:
- my_param
tasks:
task1:
action: std.echo output="Param value is <% $.my_param %>"
"""
wfs = wf_service.create_workflows(wfs_text)
self.assertEqual(2, len(wfs))
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(0, spec_parser.get_wf_definition_spec_cache_size())
wf_ex = self.engine.start_workflow('wf')
self.await_workflow_success(wf_ex.id)
# We expect to have a cache entry for every workflow execution
# but two of them should refer to the same object.
self.assertEqual(3, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(2, spec_parser.get_wf_definition_spec_cache_size())
sub_wf_execs = db_api.get_workflow_executions(name='sub_wf')
self.assertEqual(2, len(sub_wf_execs))
spec1 = spec_parser.get_workflow_spec_by_execution_id(
sub_wf_execs[0].id)
spec2 = spec_parser.get_workflow_spec_by_execution_id(
sub_wf_execs[1].id)
self.assertIs(spec1, spec2)
def test_workflow_spec_cache_update_via_workflow_service(self):
wf_text = """
version: '2.0'
wf:
tasks:
task1:
action: std.echo output="Echo"
"""
wfs = wf_service.create_workflows(wf_text)
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(0, spec_parser.get_wf_definition_spec_cache_size())
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wfs[0].id,
wfs[0].updated_at
)
self.assertEqual(1, len(wf_spec.get_tasks()))
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
# Now update workflow definition and check that cache is updated too.
wf_text = """
version: '2.0'
wf:
tasks:
task1:
action: std.echo output="1"
task2:
action: std.echo output="2"
"""
wfs = wf_service.update_workflows(wf_text)
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wfs[0].id,
wfs[0].updated_at
)
self.assertEqual(2, len(wf_spec.get_tasks()))
self.assertEqual(2, spec_parser.get_wf_definition_spec_cache_size())
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
def test_workflow_spec_cache_update_via_workflow_service(self):
wf_text = """
version: '2.0'
wf:
tasks:
task1:
action: std.echo output="Echo"
"""
wfs = wf_service.create_workflows(wf_text)
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(0, spec_parser.get_wf_definition_spec_cache_size())
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wfs[0].id, wfs[0].updated_at)
self.assertEqual(1, len(wf_spec.get_tasks()))
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
# Now update workflow definition and check that cache is updated too.
wf_text = """
version: '2.0'
wf:
tasks:
task1:
action: std.echo output="1"
task2:
action: std.echo output="2"
"""
wfs = wf_service.update_workflows(wf_text)
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wfs[0].id, wfs[0].updated_at)
self.assertEqual(2, len(wf_spec.get_tasks()))
self.assertEqual(2, spec_parser.get_wf_definition_spec_cache_size())
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
def test_workflow_spec_caching(self):
wf_text = """
version: '2.0'
wf:
tasks:
task1:
action: std.echo output="Echo"
"""
wfs = wf_service.create_workflows(wf_text)
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(0, spec_parser.get_wf_definition_spec_cache_size())
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wfs[0].id, wfs[0].updated_at)
self.assertIsNotNone(wf_spec)
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
def test_workflow_spec_caching(self):
wf_text = """
version: '2.0'
wf:
tasks:
task1:
action: std.echo output="Echo"
"""
wfs = wf_service.create_workflows(wf_text)
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(0, spec_parser.get_wf_definition_spec_cache_size())
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wfs[0].id,
wfs[0].updated_at
)
self.assertIsNotNone(wf_spec)
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
def test_cache_workflow_spec_by_execution_id(self):
wf_text = """
version: '2.0'
wf:
tasks:
task1:
action: std.echo output="Echo"
"""
wfs = wf_service.create_workflows(wf_text)
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(0, spec_parser.get_wf_definition_spec_cache_size())
wf_def = wfs[0]
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wf_def.id, wf_def.updated_at)
self.assertEqual(1, len(wf_spec.get_tasks()))
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
with db_api.transaction():
wf_ex = db_api.create_workflow_execution({
'id': '1-2-3-4',
'name': 'wf',
'workflow_id': wf_def.id,
'spec': wf_spec.to_dict(),
'state': states.RUNNING
})
# Check that we can get a valid spec by execution id.
wf_spec_by_exec_id = spec_parser.get_workflow_spec_by_execution_id(
wf_ex.id)
self.assertEqual(1, len(wf_spec_by_exec_id.get_tasks()))
# Now update workflow definition and check that cache is updated too.
wf_text = """
version: '2.0'
wf:
tasks:
task1:
action: std.echo output="1"
task2:
action: std.echo output="2"
"""
wfs = wf_service.update_workflows(wf_text)
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wfs[0].id, wfs[0].updated_at)
self.assertEqual(2, len(wf_spec.get_tasks()))
self.assertEqual(2, spec_parser.get_wf_definition_spec_cache_size())
self.assertEqual(1, spec_parser.get_wf_execution_spec_cache_size())
# Now finally update execution cache and check that we can
# get a valid spec by execution id.
spec_parser.cache_workflow_spec_by_execution_id(wf_ex.id, wf_spec)
wf_spec_by_exec_id = spec_parser.get_workflow_spec_by_execution_id(
wf_ex.id)
self.assertEqual(2, len(wf_spec_by_exec_id.get_tasks()))
def test_big_on_closures(self):
# The idea of the test is to run a workflow with a big 'on-success'
# list of tasks and big task inbound context ('task_ex.in_context)
# and observe how it influences memory consumption and performance.
# The test doesn't have any assertions related to memory(CPU) usage
# because it's quite difficult to do them. Particular metrics may
# vary from run to run and also depend on the platform.
sub_wf_text = """
version: '2.0'
sub_wf:
tasks:
task1:
action: std.noop
"""
wf_text = """
version: '2.0'
wf:
tasks:
task01:
action: std.noop
on-success: task02
task02:
action: std.test_dict size=1000 key_prefix='key' val='val'
publish:
continue_flag: true
data: <% task().result %>
on-success: task0
task0:
workflow: sub_wf
on-success: {{{__ON_SUCCESS_LIST__}}}
{{{__TASK_LIST__}}}
"""
# Generate the workflow text.
task_cnt = 50
on_success_list_str = ''
for i in range(1, task_cnt + 1):
on_success_list_str += ('\n - task{}: '
'<% $.continue_flag = true %>'.format(i))
wf_text = wf_text.replace('{{{__ON_SUCCESS_LIST__}}}',
on_success_list_str)
task_list_str = ''
task_template = """
task{}:
action: std.noop
"""
for i in range(1, task_cnt + 1):
task_list_str += task_template.format(i)
wf_text = wf_text.replace('{{{__TASK_LIST__}}}', task_list_str)
wf_service.create_workflows(sub_wf_text)
wf_service.create_workflows(wf_text)
# Start the workflow.
wf_ex = self.engine.start_workflow('wf')
self.await_workflow_success(wf_ex.id, timeout=60)
self.assertEqual(2, spec_parser.get_wf_execution_spec_cache_size())
with db_api.transaction():
wf_ex = db_api.get_workflow_execution(wf_ex.id)
task_execs = wf_ex.task_executions
self.assertEqual(task_cnt + 3, len(task_execs))
self._assert_single_item(task_execs, name='task0')
self._assert_single_item(task_execs, name='task{}'.format(task_cnt))
def test_workflow_spec_cache_update_via_workbook_service(self):
wb_text = """
version: '2.0'
name: wb
workflows:
wf:
tasks:
task1:
action: std.echo output="Echo"
"""
wb_service.create_workbook_v2(wb_text)
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(0, spec_parser.get_wf_definition_spec_cache_size())
wf = db_api.get_workflow_definition('wb.wf')
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wf.id,
wf.updated_at
)
self.assertEqual(1, len(wf_spec.get_tasks()))
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
# Now update workflow definition and check that cache is updated too.
wb_text = """
version: '2.0'
name: wb
workflows:
wf:
tasks:
task1:
action: std.echo output="1"
task2:
action: std.echo output="2"
"""
wb_service.update_workbook_v2(wb_text)
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
wf = db_api.get_workflow_definition(wf.id)
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wf.id,
wf.updated_at
)
self.assertEqual(2, len(wf_spec.get_tasks()))
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(2, spec_parser.get_wf_definition_spec_cache_size())
def test_cache_workflow_spec_by_execution_id(self):
wf_text = """
version: '2.0'
wf:
tasks:
task1:
action: std.echo output="Echo"
"""
wfs = wf_service.create_workflows(wf_text)
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(0, spec_parser.get_wf_definition_spec_cache_size())
wf_def = wfs[0]
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wf_def.id,
wf_def.updated_at
)
self.assertEqual(1, len(wf_spec.get_tasks()))
self.assertEqual(0, spec_parser.get_wf_execution_spec_cache_size())
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
wf_ex = db_api.create_workflow_execution({
'id': '1-2-3-4',
'name': 'wf',
'workflow_id': wf_def.id,
'spec': wf_spec.to_dict(),
'state': states.RUNNING
})
# Check that we can get a valid spec by execution id.
wf_spec_by_exec_id = spec_parser.get_workflow_spec_by_execution_id(
wf_ex.id
)
self.assertEqual(1, len(wf_spec_by_exec_id.get_tasks()))
# Now update workflow definition and check that cache is updated too.
wf_text = """
version: '2.0'
wf:
tasks:
task1:
action: std.echo output="1"
task2:
action: std.echo output="2"
"""
wfs = wf_service.update_workflows(wf_text)
self.assertEqual(1, spec_parser.get_wf_definition_spec_cache_size())
wf_spec = spec_parser.get_workflow_spec_by_definition_id(
wfs[0].id,
wfs[0].updated_at
)
self.assertEqual(2, len(wf_spec.get_tasks()))
self.assertEqual(2, spec_parser.get_wf_definition_spec_cache_size())
self.assertEqual(1, spec_parser.get_wf_execution_spec_cache_size())
# Now finally update execution cache and check that we can
# get a valid spec by execution id.
spec_parser.cache_workflow_spec_by_execution_id(wf_ex.id, wf_spec)
wf_spec_by_exec_id = spec_parser.get_workflow_spec_by_execution_id(
wf_ex.id
)
self.assertEqual(2, len(wf_spec_by_exec_id.get_tasks()))
def test_big_on_closures(self):
# The idea of the test is to run a workflow with a big 'on-success'
# list of tasks and big task inbound context ('task_ex.in_context)
# and observe how it influences memory consumption and performance.
# The test doesn't have any assertions related to memory(CPU) usage
# because it's quite difficult to do them. Particular metrics may
# vary from run to run and also depend on the platform.
sub_wf_text = """
version: '2.0'
sub_wf:
tasks:
task1:
action: std.noop
"""
wf_text = """
version: '2.0'
wf:
tasks:
task01:
action: std.noop
on-success: task02
task02:
action: std.test_dict size=1000 key_prefix='key' val='val'
publish:
continue_flag: true
data: <% task().result %>
on-success: task0
task0:
workflow: sub_wf
on-success: {{{__ON_SUCCESS_LIST__}}}
{{{__TASK_LIST__}}}
"""
# Generate the workflow text.
task_cnt = 50
on_success_list_str = ''
for i in range(1, task_cnt + 1):
on_success_list_str += (
'\n - task{}: '
'<% $.continue_flag = true %>'.format(i)
)
wf_text = wf_text.replace(
'{{{__ON_SUCCESS_LIST__}}}',
on_success_list_str
)
task_list_str = ''
task_template = """
task{}:
action: std.noop
"""
for i in range(1, task_cnt + 1):
task_list_str += task_template.format(i)
wf_text = wf_text.replace('{{{__TASK_LIST__}}}', task_list_str)
wf_service.create_workflows(sub_wf_text)
wf_service.create_workflows(wf_text)
# Start the workflow.
wf_ex = self.engine.start_workflow('wf')
self.await_workflow_success(wf_ex.id, timeout=60)
self.assertEqual(2, spec_parser.get_wf_execution_spec_cache_size())
with db_api.transaction():
wf_ex = db_api.get_workflow_execution(wf_ex.id)
task_execs = wf_ex.task_executions
self.assertEqual(task_cnt + 3, len(task_execs))
self._assert_single_item(task_execs, name='task0')
self._assert_single_item(task_execs, name='task{}'.format(task_cnt))