Connecting InteLIS Data¶

The InteLIS database is the laboratory data that the LLM queries against. Without it, the dashboard shows empty charts and the chat can't generate SQL queries.

There are two ways to provide InteLIS data:

Option A — Import a database dump into the Docker MySQL container
Option B — Connect to an external InteLIS instance

Option A: Import a Database Dump¶

If you have a SQL dump of the InteLIS database (e.g. from mysqldump):

1. Import the dump¶

make db-import FILE=path/to/intelis-dump.sql

Both plain .sql and compressed .sql.gz files are supported:

make db-import FILE=path/to/intelis-dump.sql.gz

This loads the file into the Docker MySQL container.

Create the database first if your dump doesn't include CREATE DATABASE

Many dumps contain only tables and data without a CREATE DATABASE or USE statement. In that case, you must create the target database before importing. You can let make db-import handle this automatically by passing the DB variable:

make db-import FILE=path/to/dump.sql DB=vlsm

This creates the database if it doesn't exist and imports into it. The DB value defaults to the QUERY_DB_NAME setting in .env (or vlsm if unset).

2. Update the database name (if needed)¶

If the dump creates a database with a name other than vlsm, update .env:

QUERY_DB_NAME=your_database_name

The database must exist in MySQL

The QUERY_DB_NAME value must match an actual database in the MySQL container. If you imported into a database called vlsm-drc, set QUERY_DB_NAME=vlsm-drc. A mismatch will cause an Unknown database error at startup.

Then restart:

make down && make up

3. Export the schema for RAG¶

The RAG system needs to know what tables and columns exist:

docker compose exec app php /var/www/bin/export-schema.php

This reads the InteLIS database schema and writes it to var/schema.json.

4. Seed the RAG index¶

Build snippets from the schema + business rules + field guide, and upload them to Qdrant:

make rag-refresh

See RAG Seeding for details on what this does.

5. Verify¶

Test that RAG can find relevant snippets:

curl -X POST http://localhost:8089/v1/search \
  -H 'Content-Type: application/json' \
  -d '{"query": "viral load suppression", "k": 3}'

Test an end-to-end query:

curl -X POST http://localhost:8080/api/v1/chat/ask \
  -H 'Content-Type: application/json' \
  -d '{"question": "How many viral load tests were done last month?"}'

Option B: Connect to an External InteLIS Instance¶

If your InteLIS database is hosted elsewhere (e.g. a shared lab server):

1. Update `.env`¶

QUERY_DB_HOST=lab-db-server.example.com
QUERY_DB_NAME=vlsm
QUERY_DB_USER=readonly_user
QUERY_DB_PASSWORD=readonly_pass

Use a read-only database user

The application only runs SELECT queries against the InteLIS database. Using a read-only MySQL user is a good security practice.

2. Restart services¶

make down && make up

3. Export the schema and seed RAG¶

docker compose exec app php /var/www/bin/export-schema.php
make rag-refresh

4. Verify¶

Same as Option A, Step 5 above.

What Changes After Connecting¶

Feature	Without InteLIS	With InteLIS
Dashboard	Loads, empty charts	Full national indicators
Chat	UI loads, AI can't query	Full conversational queries
Reports	Can be created, no data	Produce real results
RAG Index	Empty	~2,400 snippets

Updating After Schema Changes¶

If the InteLIS database schema changes (new tables, renamed columns, new data):

# Re-export the schema
docker compose exec app php /var/www/bin/export-schema.php

# Re-seed RAG with updated schema
make rag-refresh

See RAG Seeding for when to use rag-refresh vs rag-reset.