def action_over_worker_job_type(workflow_instance_id, worker_id, option, redis_score):
connection = get_connection(CONFIG.get('all'))
workflow_instance_data = _get_workflow_instance(connection, workflow_instance_id)
workflow_id = workflow_instance_data.get('type')
workflow = _get_workflow(workflow_id)
redis_key = "workflows:%s:workflow_instances:%s:workers:%s:errors" % (
workflow_instance_data.get('type'), workflow_instance_id, worker_id
)
if redis_score == 'all':
errors_list = connection.zrange(redis_key, 0, -1, withscores=True)
else:
errors_list = connection.zrangebyscore(redis_key, redis_score, redis_score, withscores=True)
if len(errors_list) > 0:
for raw_data, redis_score in errors_list:
if option == 'retry':
meta_worker = [
worker.get('meta_worker') for it_worker_id, worker in workflow.get('workers', {}).iteritems()
if it_worker_id == worker_id
]
add_data_to_next_worker(connection, meta_worker[0], raw_data)
else:
# discard
connection.zadd(
REDIS_KEY_INSTANCE_WORKER_DISCARDED % (workflow_instance_data.get('type'), workflow_instance_id, worker_id),
time.time(),
raw_data
)
connection.zremrangebyscore(redis_key, redis_score, redis_score)
else:
abort(404, "No such error id: %s" % redis_score)
return jsonify({})
def get_workflow_instance_outputs(workflow_instance_id):
connection = get_connection(CONFIG.get('all'))
workflow_instance_data = _get_workflow_instance(connection, workflow_instance_id)
workflow_id = workflow_instance_data.get('type')
workflow = _get_workflow(workflow_id)
# TODO: Manage multiple final nodes
worker_id = _find_last_worker_id(workflow, workflow.get('start_worker_id'))[0]
target_key = REDIS_KEY_INSTANCE_WORKER_FINALIZED_JOBS % (workflow_id, workflow_instance_id, worker_id)
return to_json([
json_loads(data).get('workers_output', {}).get(worker_id) for data in connection.zrange(target_key, 0, -1)
])
def get_workflow_instance_status(workflow_instance_id):
# TODO: refactor this code
# TODO: support discarded
connection = get_connection(CONFIG.get('all'))
workflow_instance_data = _get_workflow_instance(connection, workflow_instance_id)
if not workflow_instance_data.get('start_time'):
return jsonify({'status': 'pending'})
errors_keys = connection.keys(
REDIS_KEY_INSTANCE_WORKER_ERRORS % (
workflow_instance_data.get('type'), workflow_instance_id, '*'
)
)
if len(errors_keys) > 0:
return jsonify({'status': 'errors'})
total_jobs_keys = connection.keys(
REDIS_KEY_INSTANCE_WORKER_JOBS % (
workflow_instance_data.get('type'), workflow_instance_id, '*', 'total'
)
)
done_jobs_keys = connection.keys(
REDIS_KEY_INSTANCE_WORKER_JOBS % (
workflow_instance_data.get('type'), workflow_instance_id, '*', 'done'
)
)
pipe = connection.pipeline()
temp_union = 'temp:%s:%s' % (datetime.now().isoformat().split('T')[0], str(uuid.uuid4()))
pipe.sunionstore(temp_union, *total_jobs_keys)
temp_diff = 'temp:%s:%s' % (datetime.now().isoformat().split('T')[0], str(uuid.uuid4()))
pipe.sdiffstore(temp_diff, temp_union, *done_jobs_keys)
pipe.delete(temp_diff)
pipe.delete(temp_union)
execute_result = pipe.execute()
pending_jobs = execute_result[1]
if pending_jobs > 0:
status = 'running'
else:
status = 'done'
runned_for = ''
if workflow_instance_data.get('start_time') and workflow_instance_data.get('update_time'):
try:
runned_for = float(workflow_instance_data.get('update_time')) - \
float(workflow_instance_data.get('start_time'))
except:
pass
return jsonify({'status': status, 'runned_for': runned_for})
def get_workflow_instance_stats(workflow_instance_id):
connection = get_connection(CONFIG.get('all'))
workflow_instance_data = _get_workflow_instance(connection, workflow_instance_id)
workers_prefix = "workflows:%s:workflow_instances:%s:workers:" % (
workflow_instance_data.get('type'), workflow_instance_id
)
workers_data = {}
for status_type in ('done', 'total'):
pipe = connection.pipeline()
workers_keys = connection.keys(REDIS_KEY_INSTANCE_WORKER_JOBS % (workflow_instance_data.get('type'), workflow_instance_id, '*', status_type))
for key in workers_keys:
pipe.scard(key)
workers_data[status_type] = _workers_dict_data(workers_prefix, workers_keys, pipe.execute())
for status_type in ('errors', 'discarded', 'finalized'):
pipe = connection.pipeline()
workers_keys = connection.keys(workers_prefix+"*:"+status_type)
for key in workers_keys:
pipe.zcard(key)
workers_data[status_type] = _workers_dict_data(workers_prefix, workers_keys, pipe.execute())
workers_stats = {}
for status_type, data in workers_data.iteritems():
for worker, worker_num in data.iteritems():
if not workers_stats.get(worker):
workers_stats[worker] = {}
if status_type == 'jobs_ids':
workers_stats[worker]['total'] = worker_num
else:
workers_stats[worker][status_type] = worker_num
for key in connection.keys(workers_prefix+'*:timing'):
worker = key.split(workers_prefix)[1].split(':')[0]
key_values = [float(x) for x in connection.lrange(key, 0, -1)]
workers_stats[worker]['total_time'] = sum(key_values)
workers_stats[worker]['average_time'] = workers_stats[worker]['total_time']/len(key_values)
for worker, values in workers_stats.iteritems():
workers_stats[worker]['pending'] = int(values.get('total', 0)) - int(
values.get('done', 0)) - int(values.get('errors', 0)) - int(values.get('discarded', 0))
return jsonify({
'workers': workers_stats,
'start_time': workflow_instance_data.get('start_time'),
'creation_time': workflow_instance_data.get('creation_time'),
'update_time': workflow_instance_data.get('update_time')
})
def get_workflow_instance_worker_jobs_type(workflow_instance_id, worker_id, option):
connection = get_connection(CONFIG.get('all'))
workflow_instance_data = _get_workflow_instance(connection, workflow_instance_id)
jobs = connection.zrange(
"workflows:%s:workflow_instances:%s:workers:%s:%s" % (
workflow_instance_data.get('type'), workflow_instance_id, worker_id, option
),
-20, -1, withscores=True
)
for i, data in enumerate(jobs):
raw_data, score = data
jobs[i] = json.loads(raw_data)
jobs[i]['redis_score'] = score
return Response(
json.dumps(jobs), mimetype='application/json'
)
def create_workflow_workflow_instance(workflow_id):
redis_con = get_connection(CONFIG.get('all'))
workflow = _get_workflow(workflow_id)
data = request.get_json()
start_worker_id = workflow.get('start_worker_id')
meta_worker = ([
worker.get('meta_worker') for worker_id, worker in workflow.get('workers', {}).iteritems()
if worker_id == start_worker_id
] or [None])[0]
if meta_worker:
workflow_instance_id = start_workflow_instance(
redis_con, workflow_id, meta_worker, workflow.get('start_worker_id'),
data.get('name'), data.get('data', {})
)
return Response(json.dumps({'id': workflow_instance_id}), status=201)
return Response(status=404)
def get_workflow_instance(workflow_instance_id):
connection = get_connection(CONFIG.get('all'))
workflow = _get_workflow_instance(connection, workflow_instance_id)
return jsonify(workflow)
def get_workflow_workflow_instances(workflow_id):
redis_con = get_connection(CONFIG.get('all'))
workflow_instances = redis_con.zrange('workflows:%s:workflow_instances' % workflow_id, -25, -1)
return_data = [json.loads(workflow_instance) for workflow_instance in workflow_instances]
return Response(json.dumps(return_data), mimetype='application/json')