Failed Tasks

有些时候一些任务会失败。让我们看下在nornir中如何处理这些失败的任务。
让我们和之前一样从样本代码开始:

import logging

from nornir import InitNornir
from nornir.core.task import Task, Result
from nornir_utils.plugins.functions import print_result

# instantiate the nr object
nr = InitNornir(config_file="config.yaml")
# let's filter it down to simplify the output
cmh = nr.filter(site="cmh", type="host")

def count(task: Task, number: int) -> Result:
    return Result(
        host=task.host,
        result=f"{[n for n in range(0, number)]}"
    )

def say(task: Task, text: str) -> Result:
    if task.host.name == "host2.cmh":
        raise Exception("I can't say anything right now")
    return Result(
        host=task.host,
        result=f"{task.host.name} says {text}"
    )

现在用一个我们上节教程中用过的例子:

def greet_and_count(task: Task, number: int):
    task.run(
        name="Greeting is the polite thing to do",
        severity_level=logging.DEBUG,
        task=say,
        text="hi!",
    )

    task.run(
        name="Counting beans",
        task=count,
        number=number,
    )
    task.run(
        name="We should say bye too",
        severity_level=logging.DEBUG,
        task=say,
        text="bye!",
    )

    # let's inform if we counted even or odd times
    even_or_odds = "even" if number % 2 == 1 else "odd"
    return Result(
        host=task.host,
        result=f"{task.host} counted {even_or_odds} times!",
    )

大家记着我们在host2.cmh上设置了一个错误，现在我们执行这个任务看下将会发生什么:

result = cmh.run(
    task=greet_and_count,
    number=5,
)

让我们看下这个result对象:

print(result.faild)

输出:
True

print(result.faild_hosts)

输出:
{'host2.cmh': MultiResult: [Result: "greet_and_count", Result: "Greeting is the polite thing to do"]}

print(result['host2.cmh'].exception)

输出:
Subtask: Greeting is the polite thing to do (failed)

print(result['host2.cmh'][1].exception)

输出:
I can't say anything right now

如你所见，result对象是意识到了一些错误的，如果你想看的话也可以查看。
你也可以用print_result函数:

print_result(result)

输出:
greet_and_count*****************************************************************
* host1.cmh ** changed : False *************************************************
vvvv greet_and_count ** changed : False vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv INFO
host1.cmh counted even times!
---- Counting beans ** changed : False ----------------------------------------- INFO
[0, 1, 2, 3, 4]
^^^^ END greet_and_count ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
* host2.cmh ** changed : False *************************************************
vvvv greet_and_count ** changed : False vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv ERROR
Subtask: Greeting is the polite thing to do (failed)

---- Greeting is the polite thing to do ** changed : False --------------------- ERROR
Traceback (most recent call last):
  File "/home/bless/.local/lib/python3.8/site-packages/nornir/core/task.py", line 99, in start
    r = self.task(self, **self.params)
  File "/mnt/d/Project/nornir_learn2/learn_failed_tasks.py", line 20, in say
    raise Exception("I can't say anything right now")
Exception: I can't say anything right now

^^^^ END greet_and_count ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

还有一个方法可以抛出异常如果task又错误的话(目前我还没看出来这个怎么应用):

from nornir.core.exceptions import NornirExecutionError
try:
    result.raise_on_error()
except NornirExecutionError:
    print("ERROR!!!")

输出:
ERROR!!!

Skipped hosts

Norni会跟踪任务失败的hosts，并且不会在这些hosts中执行接下来的任务：

#这段代码必须要加在前面的后面才可以感受到
#host2.cmh这个host在前面的任务中失败，所以在执行这个hi时Nornir会直接跳过
def hi(task: Task) -> Result:
    return Result(host=task.host, result=f"{task.host.name}: Hi, I am still here!")

result = cmh.run(task=hi)
print_result(result)

输出:
hi******************************************************************************
* host1.cmh ** changed : False *************************************************
vvvv hi ** changed : False vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv INFO
host1.cmh: Hi, I am still here!
^^^^ END hi ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

你可以通过传递这个参数on_failed=True，来强制执行:

result = cmh.run(task=hi, on_failed=True)

print_result(result)

输出:
hi******************************************************************************
* host1.cmh ** changed : False *************************************************
vvvv hi ** changed : False vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv INFO
host1.cmh: Hi, I am still here!
^^^^ END hi ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
* host2.cmh ** changed : False *************************************************
vvvv hi ** changed : False vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv INFO
host2.cmh: Hi, I am still here!

你还可以排除一些执行成功的host,通过on_good这个参数:

result = cmh.run(task=hi, on_failed=True, on_good=False)
print_result(result)

输出:

hi******************************************************************************
* host2.cmh ** changed : False *************************************************
vvvv hi ** changed : False vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv INFO
host2.cmh: Hi, I am still here!
^^^^ END hi ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

为了实现这个，Nornir保留了一个共享的data对象，存储一系列失败的hosts

print(nr.data.failed_hosts)

输出:
{'host2.cmh'}

如果你想把一些hosts变成成功的并且使他们执行后续的任务，可以同对单独的host执行recover_host函数或者通过reset_failed_hosts重置整个列表。

nr.data.reset_failed_hosts()
print(nr.data.failed_hosts)

输出:
set()

Raise on error automatically

另外，可以通过raise_on_error配置选项来实现错误发生时自动抛出异常:

nr = InitNornir(config_file="config.yaml", core={"raise_on_error": True})
cmh = nr.filter(site="cmh", type="host")
try:
    result = cmh.run(
        task=greet_and_count,
        number=5,
    )
except NornirExecutionError:
    print("ERROR!!!")

输出:
ERROR!!!

Workflows

默认的工作流适用于大多数的用例，当错误发生时跳过这些host，并且print_result会给出足够的信息去里理解发生了什么。对于更复杂的工作流，这个框架会提供足够的空间去轻松部署。