Task instances - MaxCompute - Alibaba Cloud Documentation Center

A task, such as an SQL task, is the basic compute unit in MaxCompute. This topic describes the basic operations on task instances in PyODPS.

Basic operations

When a task is triggered, a MaxCompute instance is created for the task. For more information, see Instance. You can perform the following operations on MaxCompute instances:

list_instances(): obtains all instances in a project.
exist_instance(): checks whether an instance exists.
get_instance(): obtains an instance.
stop_instance(): stops an instance. You can stop only an instance that is running. If you attempt to stop an instance that is not running, an error is returned.
Note You can also call the stop() method to stop an instance. For example, you can run o.get_instance('instance id').stop.

Examples

Obtain the ID of an instance.

for instance in o.list_instances():
    print(instance.id)

Check whether an instance exists.

print(o.exist_instance('my_instance_id'))

Obtain the Logview URL of an instance

For an SQL task instance, you can call the get_logview_address method to obtain the Logview URL of the instance.

# Execute an SQL statement to obtain an instance. Then, call the get_logview_address() method to obtain the Logview URL of the instance. 
instance = o.run_sql('desc pyodps_iris')
print(instance.get_logview_address())

# Obtain the Logview URL of an instance that has the specified ID. 
instance = o.get_instance('my_instance_id')
print(instance.get_logview_address())

For a Machine Learning Platform for AI (PAI) instance, you must list the sub-instances of the instance and then obtain the Logview URLs of the sub-instances.

instance = o.run_xflow('AppendID', 'algo_public', {'inputTableName': 'input_table', 'outputTableName': 'output_table'})
for sub_inst_name, sub_inst in o.get_xflow_sub_instances(instance).items():
    print('%s: %s' % (sub_inst_name, sub_inst.get_logview_address()))

Obtain the status of an instance

An instance can be in the Running, Suspended, or Terminated state. You can use the following methods to obtain the status of an instance:

status: obtains the status of an instance.
is_terminated: checks whether the execution of the current instance is complete.
is_successful: checks whether the execution of the current instance is successful. If the instance is running or fails, False is returned.

Sample code:

Obtain the status of an instance.

instance=o.get_instance('my_instance_id')
print(instance.status)
print(instance.status.value)

Sample response:

Status.TERMINATED
Terminated

Check whether the execution of an instance is complete.

instance=o.get_instance('my_instance_id')
from odps.models import Instance
print(instance.status == Instance.Status.TERMINATED)

If True is returned, the execution is complete.

Note You can call the wait_for_completion method to allow the system to return the status until the execution of the instance is complete. The effect of the wait_for_success method is similar to the effect of the wait_for_completion method. However, if you call the wait_for_success method, an error is returned if the instance fails.

Perform operations on sub-instances

A running instance may contain one or more sub-instances. You can call the following methods to perform operations on sub-instances:

get_task_names: obtains all sub-instances. This method returns the names of all sub-instances.
```
instance=o.get_instance('my_instance_id')
instance.get_task_names()
```
Sample response:
```
['jdbc_sql_task']
```

get_task_result: obtains the execution result of a sub-instance. This method returns the execution result of each sub-instance in the form of a dictionary. Sample code:

Obtain the names of all sub-instances.

instance=o.execute_sql('select*frompyodps_irislimit1')
print(instance.get_task_names())

Sample response:

['AnonymousSQLTask']

Obtain the execution result of a sub-instance.

print(instance.get_task_result('AnonymousSQLTask'))

Sample response:

"sepallength","sepalwidth","petallength","petalwidth","name"
4.9,3.0,1.4,0.2,"Iris-setosa"

Obtain the execution results of sub-instances.

print(instance.get_task_results())

Sample response:

OrderedDict([('AnonymousSQLTask', '"sepallength","sepalwidth","petallength","petalwidth","name"\n4.9,3.0,1.4,0.2,"Iris-setosa"\n')])

get_task_progress: obtains the current progress of a sub-instance when the instance is running.

instance=o.get_instance('20160519101349613gzbzufck2')
while not instance.is_terminated():
    for task_name in instance.get_task_names():
        print(instance.id, instance.get_task_progress(task_name).get_stage_progress_formatted_string())
        time.sleep(10)

Sample response:

20160519101349613gzbzufck2 2016-05-19 18:14:03 M1_Stg1_job0:0/1/1[100%]