ODD Collector Profiler

Reads batch of data and uses DataProfiler for getting statistics and map them to ODD DataEntities statistics.

• ODD Collector Profiler
  • Supported data sources
  • Config example
  • Docker build
  • M1 Issue
    • Pyodbc
    • grpcio
    • tensorflow

Supported data sources

• Postgres
• Azure SQL

Config example

Key	Value
default_pulling_interval	Once per interval collector will collect statistics and send them to ODD Platform
token	Token created during collector registration via UI or programmatically
platform_host_url	ODD Platform host
profilers	List of configs for datasources profilers

collector-config.yaml

default_pulling_interval: 360
token:  <COLLECTOR_TOKEN>
platform_host_url: http://localhost:8080
profilers:
  - type: postgres
    name: my_postgres
    host: localhost
    port: 5432
    username: postgres
    password: ""
    database: db
    tables: ["some_table"]

Docker build

docker build . -t odd_collector_profiler

M1 Issue

Pyodbc

On M1 pyodbc needs to be installed as no-binary. Command below will add that info to poetry.toml:

poetry config --local installer.no-binary pyodbc

grpcio

Needs an env variables:

export GRPC_PYTHON_BUILD_SYSTEM_OPENSSL=1
export GRPC_PYTHON_BUILD_SYSTEM_ZLIB=1

tensorflow

DataProfiler uses tensorflow package for auto-labeling, there are no ready .whl for M1. Need it to be builded and used manually, read the Tensorflow documentation.