rest_api tutorial

[kjlee18]

Hi, I am new to haystack and trying to implement NLP task with haystack.
While, doing rest api tutorial, I got a problem. rest_api_tutorial_link

I follow the all instructions on the tutorial. One difference is that the current haystack version is not matched with tutorial version (current version 1.15.1, tutorial version 1.12.1)

When I do docker-compose up with docker-compose.yml, the following is happened.
The problem is that elasticsearch returns "status:420" which means too many request.
I tried to solve the problem with thread_pool.write.queue_size=2000, and thread_pool.write.size=64 but did not work.

Is there anyone experiencing the same problem or finished the problem?
Your comment will be really helpful

haystack-api_1   | 04/18/2023 10:27:49 AM Param 'host' of component 'DocumentStore' overwritten with environment variable 'DOCUMENTSTORE_PARAMS_HOST' value 'elasticsearch'.
haystack-api_1   | 04/18/2023 10:27:49 AM Param 'host' of component 'DocumentStore' overwritten with environment variable 'DOCUMENTSTORE_PARAMS_HOST' value 'elasticsearch'.
haystack-api_1   | 04/18/2023 10:27:49 AM Param 'host' of component 'DocumentStore' overwritten with environment variable 'DOCUMENTSTORE_PARAMS_HOST' value 'elasticsearch'.
haystack-api_1   | /opt/venv/lib/python3.10/site-packages/elasticsearch/connection/base.py:190: ElasticsearchDeprecationWarning: Elasticsearch built-in security features are not enabled. Without authentication, your cluster could be accessible to anyone. See https://www.elastic.co/guide/en/elasticsearch/reference/7.17/security-minimal-setup.html to enable security.
haystack-api_1   |   warnings.warn(message, category=ElasticsearchDeprecationWarning)
haystack-api_1   | /opt/venv/lib/python3.10/site-packages/elasticsearch/connection/http_urllib3.py:275: DeprecationWarning: HTTPResponse.getheaders() is deprecated and will be removed in urllib3 v2.1.0. Instead access HTTPResponse.headers directly.
haystack-api_1   |   return response.status, response.getheaders(), raw_data
haystack-api_1   | 04/18/2023 10:27:49 AM PUT http://elasticsearch:9200/document/_mapping [status:429 request:0.011s]
haystack-api_1   | 04/18/2023 10:27:49 AM Error loading query pipeline from /opt/pipelines/document-search.haystack-pipeline.yml. 
haystack-api_1   |  Failed loading pipeline component 'Retriever'. See the stacktrace above for more information.

[PAHXO]

Got it working. This is the file hierarchy.

└── rest_api_tutorial
    ├── article_txt_countries_and_capitals
    │   ├── 0_Minsk.txt
    │   ├── 0_Montenegro.txt
    │   ├── etc..
    ├── docker-compose.yml
    ├── pipelines.haystack-pipeline.yml

The docker-compose.yml file

services:

  haystack-api:
    image: "deepset/haystack:cpu"
    volumes:
      - - ./:/opt/pipelines
    ports:
      - 8000:8000
    restart: on-failure
    environment:
      - DOCUMENTSTORE_PARAMS_HOST=elasticsearch
      - PIPELINE_YAML_PATH=/opt/pipelines/pipelines.haystack-pipeline.yml
      - TOKENIZERS_PARALLELISM=false
      # Uncomment the following line to customise how much time (in seconds) a worker can spend serving a request
      # before it times out. This should include the time required to cache the models and setup the pipelines.

      # - GUNICORN_CMD_ARGS="--timeout=3000"
    depends_on:
      elasticsearch:
        condition: service_healthy

  elasticsearch:
    image: "elasticsearch:7.9.2"
    ports:
      - 9200:9200
    restart: on-failure
    environment:
      - discovery.type=single-node
      - "ES_JAVA_OPTS=-Xms1024m -Xmx1024m"
    healthcheck:
        test: curl --fail http://localhost:9200/_cat/health || exit 1
        interval: 10s
        timeout: 1s
        retries: 10

pipelines.haystack-pipeline.yml file

version: ignore

components:
  - name: DocumentStore
    type: ElasticsearchDocumentStore
    params:
      embedding_dim: 384
  - name: Retriever
    type: EmbeddingRetriever
    params:
      document_store: DocumentStore
      top_k: 10 
      embedding_model: sentence-transformers/all-MiniLM-L6-v2
  - name: FileTypeClassifier
    type: FileTypeClassifier
  - name: TextFileConverter
    type: TextConverter
  - name: Preprocessor
    type: PreProcessor
    params:
      split_by: word
      split_length: 250
      split_overlap: 30 
      split_respect_sentence_boundary: True

pipelines:
  - name: query 
    nodes:
      - name: Retriever
        inputs: [Query]
  - name: indexing
    nodes:
      - name: FileTypeClassifier
        inputs: [File]
      - name: TextFileConverter
        inputs: [FileTypeClassifier.output_1]
      - name: Preprocessor
        inputs: [TextFileConverter]
      - name: Retriever
        inputs: [Preprocessor]
      - name: DocumentStore
        inputs: [Retriever]

Just changed the docker Elasticsearch image to a newer one.

Fixed the volume directory of the docker image to match the text in the toturial.

Update or add 'ignore' the version number of the pipelines.haystack-pipeline.yml file.

[PoornaSaiNagendra]

Hi @kjlee18 and @PAHXO I am using the elasticsearch:7.9.2. I am also facing the same issues. Could you please share the steps in detail if you don't mind?

Here is my docker-compose.yml:

services:

haystack-api:
image: "deepset/haystack:cpu"
volumes:
- ./rest_api/rest_api/pipeline:/opt/pipelines
ports:
- 8000:8000
restart: on-failure
environment:
- DOCUMENTSTORE_PARAMS_HOST=elasticsearch
- PIPELINE_YAML_PATH=/opt/pipelines/pipelines.haystack-pipeline.yml
- TOKENIZERS_PARALLELISM=false
# Uncomment the following line to customise how much time (in seconds) a worker can spend serving a request
# before it times out. This should include the time required to cache the models and setup the pipelines.

  # - GUNICORN_CMD_ARGS="--timeout=3000"
depends_on:
  elasticsearch:
    condition: service_healthy

elasticsearch:
image: "elasticsearch:7.9.2"
ports:
- 9200:9200
restart: on-failure
environment:
- discovery.type=single-node
- "ES_JAVA_OPTS=-Xms1024m -Xmx1024m"
healthcheck:
test: curl --fail http://localhost:9200/_cat/health || exit 1
interval: 10s
timeout: 1s
retries: 10

[PAHXO]

I have updated my comment, you can look it up again.