[Issue #9] Setup opensearch locally (#39)

## Summary
Fixes #9

### Time to review: __10 mins__

## Changes proposed
Setup a search index to run locally via Docker

Updated makefile to automatically initialize the index + added a script
to wait for the index to start up before proceeding.

Setup a very basic client for connecting to the search index (will be
expanded more in subsequent PRs)

Basic test / test utils to verify it is working (also will be expanded)

## Context for reviewers
This is the first step in getting the search index working locally. This
actually gets it running, and the client works, we just aren't doing
anything meaningful with it yet besides tests.

## Additional information

This doesn't yet create an index that we can use, except in the test.
However, if you want to test out a search index, you can go to
http://localhost:5601/app/dev_tools#/console (after running `make init`)
to run some queries against the (one node) cluster.
https://opensearch.org/docs/latest/getting-started/communicate/#sending-requests-in-dev-tools
provides some examples of how to create + use indexes that you can
follow.

api/Makefile

@@ -100,7 +100,7 @@ start-debug:
 run-logs: start
 	docker-compose logs --follow --no-color $(APP_NAME)
 
-init: build init-db
+init: build init-db init-opensearch
 
 clean-volumes: ## Remove project docker volumes (which includes the DB state)
 	docker-compose down --volumes
@@ -179,6 +179,19 @@ create-erds: # Create ERD diagrams for our DB schema
 setup-postgres-db: ## Does any initial setup necessary for our local database to work
 	$(PY_RUN_CMD) setup-postgres-db
 
+##################################################
+# Opensearch
+##################################################
+
+init-opensearch: start-opensearch
+# TODO - in subsequent PRs, we'll add more to this command to setup the search index locally
+
+start-opensearch:
+	docker-compose up --detach opensearch-node
+	docker-compose up --detach opensearch-dashboards
+	./bin/wait-for-local-opensearch.sh
+
+
 
 ##################################################
 # Testing

api/bin/wait-for-local-opensearch.sh

@@ -0,0 +1,31 @@
+#!/bin/bash
+# wait-for-local-opensearch
+
+set -e
+
+# Color formatting
+RED='\033[0;31m'
+NO_COLOR='\033[0m'
+
+MAX_WAIT_TIME=30 # seconds
+WAIT_TIME=0
+
+# Curl the healthcheck endpoint of the local opensearch
+# until it returns a success response
+until curl --output /dev/null --silent http://localhost:9200/_cluster/health;
+do
+  echo "waiting on OpenSearch to initialize..."
+  sleep 3
+
+  WAIT_TIME=$(($WAIT_TIME+3))
+  if [ $WAIT_TIME -gt $MAX_WAIT_TIME ]
+  then
+    echo -e "${RED}ERROR: OpenSearch appears to not be starting up, running \"docker logs opensearch-node\" to troubleshoot.${NO_COLOR}"
+    docker logs opensearch-node
+    exit 1
+  fi
+done
+
+echo "OpenSearch is ready after ~${WAIT_TIME} seconds"
+
+

api/docker-compose.yml

@@ -12,6 +12,41 @@ services:
     volumes:
       - grantsdbdata:/var/lib/postgresql/data
 
+  opensearch-node:
+    image: opensearchproject/opensearch:latest
+    container_name: opensearch-node
+    environment:
+      - cluster.name=opensearch-cluster # Name the cluster
+      - node.name=opensearch-node # Name the node that will run in this container
+      - discovery.type=single-node # Nodes to look for when discovering the cluster
+      - bootstrap.memory_lock=true # Disable JVM heap memory swapping
+      - "OPENSEARCH_JAVA_OPTS=-Xms512m -Xmx512m" # Set min and max JVM heap sizes to at least 50% of system RAM
+      - DISABLE_INSTALL_DEMO_CONFIG=true # Prevents execution of bundled demo script which installs demo certificates and security configurations to OpenSearch
+      - DISABLE_SECURITY_PLUGIN=true # Disables Security plugin
+    ulimits:
+      memlock:
+        soft: -1 # Set memlock to unlimited (no soft or hard limit)
+        hard: -1
+      nofile:
+        soft: 65536 # Maximum number of open files for the opensearch user - set to at least 65536
+        hard: 65536
+    volumes:
+      - opensearch-data:/usr/share/opensearch/data # Creates volume called opensearch-data and mounts it to the container
+    ports:
+      - 9200:9200 # REST API
+      - 9600:9600 # Performance Analyzer
+
+  opensearch-dashboards:
+    image: opensearchproject/opensearch-dashboards:latest
+    container_name: opensearch-dashboards
+    ports:
+      - 5601:5601 # Map host port 5601 to container port 5601
+    expose:
+      - "5601" # Expose port 5601 for web access to OpenSearch Dashboards
+    environment:
+      - 'OPENSEARCH_HOSTS=["http://opensearch-node:9200"]'
+      - DISABLE_SECURITY_DASHBOARDS_PLUGIN=true # disables security dashboards plugin in OpenSearch Dashboards
+
   grants-api:
     build:
       context: .
@@ -28,6 +63,8 @@ services:
       - .:/api
     depends_on:
       - grants-db
+      - opensearch-node
 
 volumes:
   grantsdbdata:
+  opensearch-data:

api/local.env

@@ -59,6 +59,15 @@ DB_SSL_MODE=allow
 # could contain sensitive information.
 HIDE_SQL_PARAMETER_LOGS=TRUE
 
+############################
+# Opensearch Environment Variables
+############################
+
+OPENSEARCH_HOST=opensearch-node
+OPENSEARCH_PORT=9200
+OPENSEARCH_USE_SSL=FALSE
+OPENSEARCH_VERIFY_CERTS=FALSE
+
 ############################
 # AWS Defaults
 ############################

api/pyproject.toml

@@ -22,6 +22,7 @@ gunicorn = "^22.0.0"
 psycopg = { extras = ["binary"], version = "^3.1.10" }
 pydantic-settings = "^2.0.3"
 flask-cors = "^4.0.0"
+opensearch-py = "^2.5.0"
 
 [tool.poetry.group.dev.dependencies]
 black = "^23.9.1"
@@ -43,6 +44,12 @@ sadisplay = "0.4.9"
 ruff = "^0.4.0"
 debugpy = "^1.8.1"
 freezegun = "^1.5.0"
+# This isn't the latest version of types-requests
+# because otherwise it depends on urllib3 v2 but opensearch-py
+# needs urlib3 v1. This should be temporary as opensearch-py
+# has an unreleased change to switch to v2, so I'm guessing
+# in the next few weeks we can just make this the latest?
+types-requests = "2.31.0.1"
 
 [build-system]
 requires = ["poetry-core>=1.0.0"]

api/src/adapters/search/__init__.py

@@ -0,0 +1,4 @@
+from src.adapters.search.opensearch_client import SearchClient, get_opensearch_client
+from src.adapters.search.opensearch_config import get_opensearch_config
+
+__all__ = ["SearchClient", "get_opensearch_client", "get_opensearch_config"]

api/src/adapters/search/opensearch_client.py

@@ -0,0 +1,36 @@
+from typing import Any
+
+import opensearchpy
+
+from src.adapters.search.opensearch_config import OpensearchConfig, get_opensearch_config
+
+# More configuration/setup coming in:
+# TODO - https://github.com/navapbc/simpler-grants-gov/issues/13
+
+# Alias the OpenSearch client so that it doesn't need to be imported everywhere
+# and to make it clear it's a client
+SearchClient = opensearchpy.OpenSearch
+
+
+def get_opensearch_client(
+    opensearch_config: OpensearchConfig | None = None,
+) -> SearchClient:
+    if opensearch_config is None:
+        opensearch_config = get_opensearch_config()
+
+    # See: https://opensearch.org/docs/latest/clients/python-low-level/ for more details
+    return opensearchpy.OpenSearch(**_get_connection_parameters(opensearch_config))
+
+
+def _get_connection_parameters(opensearch_config: OpensearchConfig) -> dict[str, Any]:
+    # TODO - we'll want to add the AWS connection params here when we set that up
+    # See: https://opensearch.org/docs/latest/clients/python-low-level/#connecting-to-amazon-opensearch-serverless
+
+    return dict(
+        hosts=[{"host": opensearch_config.host, "port": opensearch_config.port}],
+        http_compress=True,
+        use_ssl=opensearch_config.use_ssl,
+        verify_certs=opensearch_config.verify_certs,
+        ssl_assert_hostname=False,
+        ssl_show_warn=False,
+    )

api/src/adapters/search/opensearch_config.py

@@ -0,0 +1,33 @@
+import logging
+
+from pydantic import Field
+from pydantic_settings import SettingsConfigDict
+
+from src.util.env_config import PydanticBaseEnvConfig
+
+logger = logging.getLogger(__name__)
+
+
+class OpensearchConfig(PydanticBaseEnvConfig):
+    model_config = SettingsConfigDict(env_prefix="OPENSEARCH_")
+
+    host: str  # OPENSEARCH_HOST
+    port: int  # OPENSEARCH_PORT
+    use_ssl: bool = Field(default=True)  # OPENSEARCH_USE_SSL
+    verify_certs: bool = Field(default=True)  # OPENSEARCH_VERIFY_CERTS
+
+
+def get_opensearch_config() -> OpensearchConfig:
+    opensearch_config = OpensearchConfig()
+
+    logger.info(
+        "Constructed opensearch configuration",
+        extra={
+            "host": opensearch_config.host,
+            "port": opensearch_config.port,
+            "use_ssl": opensearch_config.use_ssl,
+            "verify_certs": opensearch_config.verify_certs,
+        },
+    )
+
+    return opensearch_config

api/tests/conftest.py

@@ -11,6 +11,7 @@
 import src.adapters.db as db
 import src.app as app_entry
 import tests.src.db.models.factories as factories
+from src.adapters import search
 from src.constants.schema import Schemas
 from src.db import models
 from src.db.models.lookup.sync_lookup_values import sync_lookup_values
@@ -143,6 +144,34 @@ def test_foreign_schema(db_schema_prefix):
     return f"{db_schema_prefix}{Schemas.LEGACY}"
 
 
+####################
+# Opensearch Fixtures
+####################
+
+
+@pytest.fixture(scope="session")
+def search_client() -> search.SearchClient:
+    return search.get_opensearch_client()
+
+
+@pytest.fixture(scope="session")
+def opportunity_index(search_client):
+    # TODO - will adjust this in the future to use utils we'll build
+    # for setting up / aliasing indexes. For now, keep it simple
+
+    # create a random index name just to make sure it won't ever conflict
+    # with an actual one, similar to how we create schemas for database tests
+    index_name = f"test_{uuid.uuid4().int}_opportunity"
+
+    search_client.indices.create(index_name, body={})
+
+    try:
+        yield index_name
+    finally:
+        # Try to clean up the index at the end
+        search_client.indices.delete(index_name)
+
+
 ####################
 # Test App & Client
 ####################