Pinecone Local is an in-memory Pinecone Vector Database emulator available as a Docker image. It provides developers with a powerful tool for local development and testing.
It integrates smoothly into CI/CD environments, allowing efficient and cost-effective testing without a live billing account.
In this article, we’ll explore how you can use Pinecone Local in your GitHub Actions workflows to do API contract testing, reduce costs and speed up your CICD testing jobs.
You can use GitHub Actions and Pinecone Local to build the following workflow, which you can configure to run whenever changes are pushed on a feature branch, or merged to main:
name: Pinecone CI/CD with Local
on:
push:
branches: [ main ]
pull_request:
branches: [ main ]
jobs:
test:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v2
- name: Set up Python
uses: actions/setup-python@v2
with:
python-version: '3.x'
- name: Set up Docker
uses: docker-practice/actions-setup-docker@master
- name: Start Pinecone Local
run: |
docker pull ghcr.io/pinecone-io/pinecone-index:latest
docker run -d \
--name pinecone-local \
-e PORT=5081 \
-e INDEX_TYPE=serverless \
-e DIMENSION=768 \
-e METRIC=cosine \
-p 5081:5081 \
--platform linux/amd64 \
ghcr.io/pinecone-io/pinecone-index:latest
- name: Install dependencies
run: |
python -m pip install --upgrade pip
pip install "pinecone[grpc]" pytest
- name: Run tests
env:
PINECONE_API_KEY: dummy-key
PINECONE_ENVIRONMENT: local
PINECONE_INDEX: my-index
run: |
pytest tests/
- name: Stop Pinecone Local
run: docker stop pinecone-localLet's look at a practical example of writing some Python code to run against our Pinecone Local instance.
Next, we run an instance of Pinecone Local, using environment variables to configure its functionality and the port it will listen on:
# Start Pinecone Local with one index - take note of the port mappings
docker run -d \
--name index1 \
-e PORT=5081 \
-e INDEX_TYPE=serverless \
-e DIMENSION=2 \
-e METRIC=cosine \
-p 5081:5081 \
--platform linux/amd64 \
ghcr.io/pinecone-io/pinecone-index:latestfrom pinecone.grpc import PineconeGRPC, GRPCClientConfig
import time
# Initialize a client. An API key must be passed, but the
# value does not matter.
pc = PineconeGRPC(api_key="pclocal")
# Target the indexes. Use the host and port number and disable TLS (SSL)
# connections since we're going over localhost
index1 = pc.Index(host="localhost:5081", grpc_config=GRPCClientConfig(secure=False))
# Upsert records into index1
index1.upsert(
vectors=[
{
"id": "vec1",
"values": [1.0, 1.5],
"metadata": {"genre": "comedy"}
},
{
"id": "vec2",
"values": [2.0, 1.0],
"metadata": {"genre": "drama"}
},
{
"id": "vec3",
"values": [0.1, 3.0],
"metadata": {"genre": "comedy"}
}
],
namespace="example-namespace"
)
# Wait for the indexes to be updated
time.sleep(5)
# Check the number of records in each index
print(index1.describe_index_stats())
# Query index2 with a metadata filter
query = index1.query(
vector=[1.0, 1.5],
filter={"genre": {"$eq": "comedy"}},
top_k=1,
include_values=True,
include_metadata=True,
namespace='example-namespace'
)
print(query)# Output of describe_index_stats call
{'dimension': 2,
'index_fullness': 0.0,
'namespaces': {'example-namespace': {'vector_count': 3}},
'total_vector_count': 3}
# Output of query
{'matches': [{'id': 'vec1',
'metadata': {'genre': 'comedy'},
'score': 1.0,
'sparse_values': {'indices': [], 'values': []},
'values': [1.0, 1.5]}],
'namespace': 'example-namespace'}
Pinecone Local offers a powerful solution for integrating vector database testing into CI/CD pipelines. Providing a containerized, in-memory emulator of Pinecone's vector database enables faster, more reliable, and cost-effective testing processes.
Pinecone Local can streamline your development workflow and make building comprehensive test coverage for your projects easier.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。