Migrate mimir-rag from Supabase to generic Postgres with Prisma

Author: smutyala1at

mimir-rag/Dockerfile

@@ -2,17 +2,27 @@
 
 FROM node:20-alpine AS builder
 WORKDIR /app
+
+# Install Python and build tools needed for tree-sitter native modules
+RUN apk add --no-cache python3 make g++
+
 COPY package*.json ./
 COPY tsconfig.json ./
+COPY prisma.config.ts ./
 COPY prisma ./prisma
 COPY src ./src
-RUN npm ci --omit=optional && \
-    npx prisma generate && \
-    npm run build && \
-    npm prune --omit=dev --omit=optional && \
-    # Remove unnecessary files from node_modules
-    find node_modules -type f \( -name "*.md" -o -name "*.ts" -o -name "*.map" \) -delete && \
-    find node_modules -type d \( -name "test" -o -name "tests" -o -name "__tests__" -o -name "docs" \) -exec rm -rf {} + 2>/dev/null || true
+
+RUN npm ci --omit=optional
+RUN PRISMA_SKIP_DATABASE_URL_CHECK=true npx prisma generate
+RUN npm run build || (echo "Build failed!" && exit 1)
+RUN ls -la dist/ || (echo "dist directory not found after build" && exit 1)
+RUN npm prune --omit=dev --omit=optional
+
+# Clean up build tools to reduce image size (they're no longer needed after npm install)
+RUN apk del python3 make g++
+
+RUN find node_modules -type f \( -name "*.md" -o -name "*.ts" -o -name "*.map" \) -delete 2>/dev/null || true
+RUN find node_modules -type d \( -name "test" -o -name "tests" -o -name "__tests__" -o -name "docs" \) -exec rm -rf {} + 2>/dev/null || true
 
 FROM node:20-alpine AS runner
 WORKDIR /app
@@ -26,10 +36,14 @@ RUN apk add --no-cache postgresql-client && \
 COPY --from=builder /app/node_modules ./node_modules
 COPY --from=builder /app/dist ./dist
 COPY --from=builder /app/package.json ./package.json
+COPY --from=builder /app/prisma.config.ts ./
 COPY --from=builder /app/prisma ./prisma
 COPY .env.example ./
 COPY scripts ./scripts
 
+# Make entrypoint script executable
+RUN chmod +x ./scripts/docker-entrypoint.sh
+
 # Create non-root user and switch to it
 RUN addgroup -g 1001 -S nodejs && \
     adduser -S nodejs -u 1001 && \

mimir-rag/Makefile

@@ -59,7 +59,7 @@ baseline-db:
 	echo "Baselining existing database schema (marking migration as applied)..."; \
 	npx prisma migrate resolve --applied 0_init && echo "Baseline complete! You can now run 'make setup-db' or 'make server'." || echo "Note: Migration may already be marked as applied."
 
-server:
+server: setup-db
 	@if [ -f "$(CONFIG_PATH)" ]; then \
 		set -a; \
 		. "$(CONFIG_PATH)"; \

mimir-rag/README.md

@@ -41,6 +41,7 @@ The database schema is automatically initialized when you run `make server` if y
 Provide the full PostgreSQL connection string in `.env`:
 ```bash
 MIMIR_DATABASE_URL=postgresql://user:password@host:5432/database
+# For Supabase: Use Session Pooler (port 6543) if using Docker - see README for details
 ```
 
 **Important:** The default schema uses `vector(3072)` for embeddings. If your embedding model uses a different dimension, you must update the database schema:
@@ -141,6 +142,8 @@ Key configuration variables include:
 - **Database**: `MIMIR_DATABASE_URL` (required) - PostgreSQL connection string
 
   **Note:** Update the embedding dimension in the database schema (`vector(3072)`) to match your embedding model's output dimension.
+  
+  **For Supabase users with Docker:** If your Supabase database shows "Not IPv4 compatible", use the **Session Pooler** connection string instead of the direct connection. Get it from: Supabase Dashboard → Settings → Database → Connection Pooling → Session mode (port 6543). This provides IPv4 compatibility required for Docker.
 - **GitHub** (language-agnostic code + docs ingestion): 
   - `MIMIR_GITHUB_URL` - Main repository URL (fallback if separate repos not set)
   - `MIMIR_GITHUB_CODE_URL` - Separate repository for code (TypeScript, Python, etc.) (optional)

mimir-rag/package.json

@@ -25,10 +25,10 @@
     "@types/node": "^24.10.0",
     "@types/pg": "^8.11.10",
     "pino-pretty": "^13.1.2",
-    "prisma": "^7.0.0",
     "tsx": "^4.20.6"
   },
   "dependencies": {
+    "prisma": "^7.0.0",
     "@ai-sdk/anthropic": "^2.0.50",
     "@ai-sdk/google": "^2.0.44",
     "@ai-sdk/mistral": "^2.0.25",

mimir-rag/prisma.config.ts

@@ -1,8 +1,29 @@
+/// <reference types="node" />
 import "dotenv/config";
 import { defineConfig } from "prisma/config";
-import { getDatabaseUrl } from "./src/utils/getDatabaseUrl";
 
-const databaseUrl = getDatabaseUrl();
+function getDatabaseUrl(): string {
+    if (process.env.DATABASE_URL) {
+        return process.env.DATABASE_URL;
+    }
+    if (process.env.MIMIR_DATABASE_URL) {
+        return process.env.MIMIR_DATABASE_URL;
+    }
+    throw new Error("DATABASE_URL or MIMIR_DATABASE_URL must be set");
+}
+
+function getDatabaseUrlSafe(): string {
+    if (process.env.PRISMA_SKIP_DATABASE_URL_CHECK === "true") {
+        return process.env.DATABASE_URL || process.env.MIMIR_DATABASE_URL || "postgresql://placeholder";
+    }
+    try {
+        return getDatabaseUrl();
+    } catch (error) {
+        throw error;
+    }
+}
+
+const databaseUrl = getDatabaseUrlSafe();
 
 export default defineConfig({
     schema: "prisma/schema.prisma",

mimir-rag/src/config/loadConfig.ts

@@ -1,6 +1,7 @@
 import { config as loadDotenv } from "dotenv";
 import path from "node:path";
 import type { AppConfig, LLMProviderName, CodeRepoConfig, DocsRepoConfig } from "./types";
+import { getDatabaseUrl } from "../utils/getDatabaseUrl";
 
 const PACKAGE_ROOT = path.resolve(__dirname, "..", "..");
 
@@ -157,7 +158,7 @@ export async function loadAppConfig(configPath?: string): Promise<AppConfig> {
         throw new Error("Server configuration must include MIMIR_SERVER_API_KEY.");
     }
 
-    const databaseUrl = getEnv("MIMIR_DATABASE_URL");
+    const databaseUrl = getDatabaseUrl();
     const databaseTable = getEnv("MIMIR_DATABASE_TABLE", false) ?? "docs";
 
     const embeddingProvider = getEnv("MIMIR_LLM_EMBEDDING_PROVIDER") as LLMProviderName;

mimir-rag/src/database/chunks.ts

@@ -86,7 +86,7 @@ export async function upsertChunks(
         const chunkValues: any[] = [
             chunk.content,
             chunk.contextualText,
-            `[${chunk.embedding.join(",")}]`, // Format as PostgreSQL array for vector type
+            `[${chunk.embedding.join(",")}]`,
             chunk.filepath,
             chunk.chunkId,
             chunk.chunkTitle,
@@ -99,7 +99,13 @@ export async function upsertChunks(
             chunk.startLine ?? null,
             chunk.endLine ?? null
         ];
-        const chunkPlaceholders = chunkValues.map((_, i) => `$${values.length + i + 1}`).join(", ");
+        const chunkPlaceholders = chunkValues.map((_, i) => {
+            const paramIndex = values.length + i + 1;
+            if (i === 2) {
+                return `$${paramIndex}::vector`;
+            }
+            return `$${paramIndex}`;
+        }).join(", ");
         placeholders.push(`(${chunkPlaceholders})`);
         values.push(...chunkValues);
     });

mimir-rag/src/database/search.ts

@@ -11,10 +11,10 @@ export async function matchDocuments(
     similarityThreshold?: number
 ): Promise<RetrievedChunk[]> {
     const threshold = similarityThreshold ?? 0.75;
-    const embeddingArray = `[${embedding.join(",")}]`;
+    const embeddingString = `[${embedding.join(",")}]`;
     const result = await pool.query(
         `SELECT * FROM match_docs($1::vector, $2, $3)`,
-        [embeddingArray, matchCount, threshold]
+        [embeddingString, matchCount, threshold]
     );
 
     return result.rows.map((row: PostgresDocRow) => ({