Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Cannot upload model to rasa x 1.1 #289

Open
nyejon opened this issue Apr 19, 2022 · 12 comments
Open

Cannot upload model to rasa x 1.1 #289

nyejon opened this issue Apr 19, 2022 · 12 comments

Comments

@nyejon
Copy link

nyejon commented Apr 19, 2022 

    return await await_and_return_response(args, kwargs, request)
  File "/usr/local/lib/python3.8/dist-packages/rasax/community/api/decorators.py", line 176, in await_and_return_response
    response = await response
  File "/usr/local/lib/python3.8/dist-packages/rasax/community/api/decorators.py", line 84, in decorated_function
    return await f(request, user=user, *args, **kwargs)
  File "/usr/local/lib/python3.8/dist-packages/rasax/community/api/blueprints/models.py", line 69, in upload_model
    minimum_version = await model_service.minimum_compatible_version()
  File "/usr/local/lib/python3.8/dist-packages/rasax/community/services/model_service.py", line 208, in minimum_compatible_version
    info = await stack_service.version()
AttributeError: 'NoneType' object has no attribute 'version'
INFO:sanic.access:
@willymwai
Copy link

Was facing a similar problem, I solved it by enabling rasaProduction in the helm chart. It's false by default. When enabled, it adds a production url to environments.yml

@sara-tagger
Copy link

Thanks for raising this issue, @hsm207 will get back to you about it soon✨

Please also check out the docs and the forum in case your issue was raised there too 🤗

@hsm207 hsm207 closed this as completed Apr 20, 2022
@nyejon
Copy link
Author

nyejon commented Apr 20, 2022

Why was this closed?

rasa-x-helm/charts/rasa-x/values.yaml

Lines 223 to 235 in 68b3682

rasaProduction:
# enable the rasa-production deployment
# You can disable the rasa-production deployment in order to use external Rasa OSS deployment instead.
enabled: false
# Define if external Rasa OSS should be used.
external:
# enable external Rasa OSS
enabled: false
# url of external Rasa OSS deployment
url: "http://rasa-bot"

States that we should be able to disable the rasa production deployment if we use an external server. I have populated the external server url, and set it to true.

I am trying to use the rasa-oss chart to deploy the models as suggested, but so much seems broken/inconsistent between them.

@nyejon
Copy link
Author

nyejon commented Apr 20, 2022 •
edited
Loading

From the helm chart:

# rasa: Settings common for all Rasa containers
# deprecated: the Rasa OSS deployment is deprecated and will be removed in the feature
# from this chart.
# It's recommended to use the rasa helm chart instead.
# see: https://github.com/RasaHQ/helm-charts/tree/main/charts/rasa#quick-start

@hsm207
Copy link

hsm207 commented Apr 20, 2022

I am trying to use the rasa-oss chart to deploy the models as suggested, but so much seems broken/inconsistent between them.

@virtualroot could you please look into this?

@hsm207 hsm207 reopened this Apr 20, 2022
@virtualroot
Copy link
Contributor

@nyejon Rasa Enterprise 1.1 is no longer compatible with Rasa Open Source 2.
Is it possible the model was made with Rasa 2.x?

@nyejon
Copy link
Author

nyejon commented Apr 20, 2022

No, I am trying to upload a freshly trained model with rasa 3.0.9.

@nyejon
Copy link
Author

nyejon commented Apr 20, 2022

image

Nothing is populated in the environments as well.

@nyejon
Copy link
Author

nyejon commented Apr 20, 2022 •
edited
Loading

I would guess there is something wrong with this line:

rasa-x-helm/charts/rasa-x/templates/_rasa.tpl

Line 3 in 68b3682

{{- if and (not .Values.rasa.versions.rasaProduction.enabled) .Values.rasa.versions.rasaProduction.external.enabled (not (empty .Values.rasa.versions.rasaProduction.external.url)) (not ((index .Values "rasa-bot").subchart)) }}

Also, with the logic from the charts, when using rasa-x as the config endpoint, should all the values for the rasa-oss deployment be populated in the rasa-x helm chart or the rasa-oss helm chart?

For example, the tracker-database name -> According to the rasa-X chart, the rasa config is deprecated, but according to the rasa-oss chart, the config is pulled from rasa-x

https://github.com/RasaHQ/helm-charts/blob/fb898a2613a4073b2a0233891df62e7db1e2da1a/charts/rasa/values.yaml#L59

This would apply to the rasa worker too. I don't see a way to specify from the Rasa OSS deployment, if it should pull production config from rasa x or the worker config.

@nyejon
Copy link
Author

nyejon commented Apr 20, 2022

ok, I seem to have gotten something more working. It now populates the environments if I enable:
rasaWorker:
external:
enabled: true

@tczekajlo
Copy link
Contributor

I would guess there is something wrong with this line:

rasa-x-helm/charts/rasa-x/templates/_rasa.tpl

Line 3 in 68b3682

{{- if and (not .Values.rasa.versions.rasaProduction.enabled) .Values.rasa.versions.rasaProduction.external.enabled (not (empty .Values.rasa.versions.rasaProduction.external.url)) (not ((index .Values "rasa-bot").subchart)) }}

Also, with the logic from the charts, when using rasa-x as the config endpoint, should all the values for the rasa-oss deployment be populated in the rasa-x helm chart or the rasa-oss helm chart?

No, using Rasa X as a config endpoint for Rasa OSS makes that Rasa OSS knows about Tracker Store, Lock Store, and so on, but it doesn't mean that Rasa X knows about Rasa OSS.

External deployments for the production or worker environment have to be defined in the values for Rasa X.

Example:

rasa:
  token: rasaToken
  versions:
    rasaProduction:
      enabled: false
      external:
        enabled: true
        url: "http://rasa:5005" # address of an external Rasa OSS deployment

@nyejon
Copy link
Author

nyejon commented Apr 20, 2022 •
edited
Loading

External deployments for the production or worker environment have to be defined in the values for Rasa X.

Example:

I have all of that configured. How would I let a worker know which tracker database to use?

I can now get the OSS deployment to try and load a model I uploaded and was trained with 3.0.9.
This results in the following error:

  File "/opt/venv/lib/python3.8/site-packages/rasa/engine/graph.py", line 390, in _load_component
    self._component: GraphComponent = constructor(  # type: ignore[no-redef]
  File "/opt/venv/lib/python3.8/site-packages/rasa/nlu/classifiers/diet_classifier.py", line 1089, in load
    return cls._load(
  File "/opt/venv/lib/python3.8/site-packages/rasa/nlu/classifiers/diet_classifier.py", line 1115, in _load
    ) = cls._load_from_files(model_path)
  File "/opt/venv/lib/python3.8/site-packages/rasa/nlu/classifiers/diet_classifier.py", line 1152, in _load_from_files
    data_example = io_utils.pickle_load(
  File "/opt/venv/lib/python3.8/site-packages/rasa/utils/io.py", line 104, in pickle_load
    return pickle.load(f)
ModuleNotFoundError: No module named 'scipy.sparse._coo'
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
  File "/opt/venv/lib/python3.8/site-packages/rasa/core/agent.py", line 86, in _update_model_from_server
    _load_and_set_updated_model(agent, temporary_directory, new_fingerprint)
  File "/opt/venv/lib/python3.8/site-packages/rasa/core/agent.py", line 69, in _load_and_set_updated_model
    agent.load_model(model_directory, fingerprint)
  File "/opt/venv/lib/python3.8/site-packages/rasa/core/agent.py", line 352, in load_model
    self.processor = MessageProcessor(
  File "/opt/venv/lib/python3.8/site-packages/rasa/core/processor.py", line 96, in __init__
    self.model_filename, self.model_metadata, self.graph_runner = self._load_model(
  File "/opt/venv/lib/python3.8/site-packages/rasa/core/processor.py", line 121, in _load_model
    metadata, runner = loader.load_predict_graph_runner(
  File "/opt/venv/lib/python3.8/site-packages/rasa/engine/loader.py", line 29, in load_predict_graph_runner
    runner = graph_runner_class.create(
  File "/opt/venv/lib/python3.8/site-packages/rasa/engine/runner/dask.py", line 51, in create
    return cls(graph_schema, model_storage, execution_context, hooks)
  File "/opt/venv/lib/python3.8/site-packages/rasa/engine/runner/dask.py", line 37, in __init__
    self._instantiated_nodes: Dict[Text, GraphNode] = self._instantiate_nodes(
  File "/opt/venv/lib/python3.8/site-packages/rasa/engine/runner/dask.py", line 60, in _instantiate_nodes
    return {
  File "/opt/venv/lib/python3.8/site-packages/rasa/engine/runner/dask.py", line 61, in <dictcomp>
    node_name: GraphNode.from_schema_node(
  File "/opt/venv/lib/python3.8/site-packages/rasa/engine/graph.py", line 528, in from_schema_node
    return cls(
  File "/opt/venv/lib/python3.8/site-packages/rasa/engine/graph.py", line 379, in __init__
    self._load_component()
  File "/opt/venv/lib/python3.8/site-packages/rasa/engine/graph.py", line 402, in _load_component
    raise GraphComponentException(
rasa.engine.exceptions.GraphComponentException: Error initializing graph component for node 'run_DIETClassifier7'.

Edit, found this: RasaHQ/rasa#10908

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
6 participants
@virtualroot @hsm207 @willymwai @tczekajlo @nyejon @sara-tagger