Question: how to save state to disk about python-statemachine HOT 5 CLOSED

Hi @UweSauter, I've wrote a small example of how this can be done using a custom model.

Please note that I've implemented this as issue358_persistence.md example file, so the tests actually run locally (the getfixture is a fixture from pytest to get a temporary file)

Issue 358

A StateMachine that has its current state persisted automatically using a custom model.

Given a generic API from a model with a state property: GenericStorageModel:

>>> from abc import ABC

>>> from abc import abstractmethod

>>> class GenericStorageModel(ABC):
...     """Abstract Base Class for persistent models.
...
...     Sub classes should implement concrete strategies for:
...
...     - `_read_state`: Read the state from the concrete persistent layer.
...     - `_write_state`: Write the state from the concrete persistent layer.
...     """
...     def __init__(self,):
...         self._state = None
...
...     def __repr__(self):
...         return f"{type(self).__name__}(state={self.state})"
...
...     @property
...     def state(self):
...         if self._state is None:
...             self._state = self._read_state()
...         return self._state
...
...     @state.setter
...     def state(self, value):
...         self._state = value
...         self._write_state(value)
...
...     @abstractmethod
...     def _read_state(self):
...         raise NotImplementedError()
...
...     @abstractmethod
...     def _write_state(self, value):
...         raise NotImplementedError()

You can write many concrete implementations of a storage strategy, persisting to YAML, JSON,
database, etc. For this example to be complete, I have for you here a reference on how to write into a file.

>>> class FileStorageModel(GenericStorageModel):
...     """A concrete implementation of a storage strategy for a Model"""
...     def __init__(self, file_path):
...         super().__init__()
...         self.file_path = file_path
...
...     def _read_state(self):
...         if not self.file_path.exists():
...             return
...
...         with self.file_path.open('r') as f:
...             return f.read().strip()
...
...     def _write_state(self, value):
...         with self.file_path.open('w+') as f:
...             return f.write(value)

You can use this FileStorageModel or any other persistent strategy that you like as a domain
model.

So, given this simple SM:

>>> from statemachine import StateMachine, State

>>> class ResourceManagement(StateMachine):
...     power_off = State(initial=True)
...     power_on = State()
...
...     turn_on = power_off.to(power_on)

We can create instances:

>>> tmp_path = getfixture('tmp_path')

>>> state_path = tmp_path / 'resource_state.txt'

>>> model = FileStorageModel(file_path=state_path)

>>> sm = ResourceManagement(model=model)

>>> sm.current_state.id
'power_off'

>>> sm.send("turn_on")

>>> sm.current_state.id
'power_on'

>>> del sm

>>> del model

Restore the previous state from disk:

>>> model = FileStorageModel(file_path=state_path)

>>> sm = ResourceManagement(model=model)

>>> sm.current_state.id
'power_on'

from python-statemachine.

Wow, thank you. I didn't expect to get such a detailed example.

I'll need to start coding to see if this fits my requirements.

Thanks again.

from python-statemachine.

Hi @UweSauter , how are you?

You can inspect the current state using the current_state property of the state machine, or in this case, you can also use the low level current_state_value property, that can be assigned to any valid state value completely bypassing all the hooks and validation. This way you can read/write the current state.

But there's no built-in function for serialization, and you can use plain standard Python to read current_state_value and write to disk.

As suggestion, you can implement a generic on_enter_state hook that does this at every state change.

Let me know if you need any help on this, if so, consider adding more concrete info about how is your use case and how do you plan to store the state.

Best regards!

from python-statemachine.

Hi Fernando,

thanks for the explanation.

So, I'm at the very early phase of making a concept for a compute resource management software. The software should track the various states the compute servers can have and transition between these states upon internal and external triggers.

An external trigger would be that the administrator wants to turn on power and the management software then calls ipmitool power on for that particular server (and transition the state from power_off to power_on).
An internal trigger would be the repeating health checks that are executed by the management software and upon success (or failure) of the check, the state is changed accordingly.

As suggestion, you can implement a generic on_enter_state hook that does this at every state change.

This way it would be able to store the state on disk, be it as JSON or YAML or in a database.
What I don't get is: how could I restore the states from disk when the management software is started (e.g. after a reboot)?

Would it also be possible to pickle.dumps() the state machine instance and restore that state machine later with pickle.loads()?

Regards,

Uwe

from python-statemachine.

@UweSauter I will close the issue for now, feel free to comment or open an issue if something is missing, ok?

Best regards!

from python-statemachine.