mcp-persist: Production-grade durable EventStore backends + zero-config proxy for MCP servers

[Ar-maan05]

Pre-submission Checklist

• This post follows the MCP community guidelines

What would you like to share?

mcp-persist is now at v1.7 — drop-in durable EventStore backends for SQLite, Redis, and PostgreSQL that survive process restarts and scale across multi-worker deployments, keeping SSE stream resumability intact.

The Problem

When an MCP client reconnects, the server has to replay the events it missed — and with only the SDK's in-memory EventStore, that replay is impossible: the session lived in one process's memory, so a restart or a reconnect to a different worker loses it.

What mcp-persist gives you

1. One-liner FastMCP integration

from mcp.server.fastmcp import FastMCP
from mcp_persist import with_persistence

mcp = FastMCP(name="MyServer")
app = with_persistence(mcp, backend="sqlite", url="events.db", ttl=3600)
# → runnable Starlette ASGI app with full resumability

2. PersistenceProxy — resumability for any MCP server
Run a lightweight proxy in front of black-box, third-party, or non-Python servers. No changes needed on the upstream or client side.
CLI example:

mcp-persist-proxy --upstream http://localhost:8001 \
    --backend sqlite --url events.db --port 8000

3. Three mature backends

SQLiteEventStore — zero external services, perfect for single-process / local / edge
RedisEventStore — multi-worker / load-balanced deployments
PostgresEventStore — team-scale, multi-node, durable

Extra production features

• Write-behind commits (SQLite)
• Payload compression
• Batched purge + PurgeScheduler
• Real-time subscribe() streaming (Redis pub/sub, Postgres LISTEN/NOTIFY, SQLite polling)
• Migration tooling between backends
• Metrics hooks
• Comprehensive tests + benchmarks

Relevant Links

https://github.com/Ar-maan05/mcp-persist