Coding Python Rpyc

RPyC -- Remote Python Call

RPyC is a transparent, symmetric Python library for remote procedure calls, clustering, and distributed computing. Object-proxying makes remote objects behave like local ones.

Use the Read tool to load referenced files identified as relevant for full details.

Install: pip install rpyc Default ports: Classic 18812, SSL 18821, Registry 18811 Python: CPython 3.8+ (rpyc 6.x Requires-Python >=3.8; no Python 2<->3 crossing) Dependencies: plumbum is unconditional (rpyc 6.x declares Requires-Dist: plumbum, no extra, no marker), so pip install rpyc always installs it; pywin32 is optional, needed only for PipeStream on Windows Repository: https://github.com/tomerfiliba-org/rpyc

When to Use

• Remote testing -- run tests centrally, operations happen on remote machines
• Administration -- control heterogeneous machines from one place using Python
• Remote hardware -- access ctypes, /dev files, drivers on remote machines transparently
• Parallel execution -- overcome GIL by distributing work across RPyC processes
• Distributed computation -- platform-agnostic foundation for clustering
• Remote services -- implement secure RPC services without heavyweight frameworks
• Monkey-patching -- replace local modules with remote ones to cross network boundaries

When NOT to Use

• Need a REST/HTTP API (use Flask, FastAPI)
• Need language-agnostic RPC (use gRPC, Thrift)
• Need message queuing (use Celery, RabbitMQ)
• Untrusted clients over the Internet without SSL/SSH wrapping

Which File Do I Need?

Reference files (distilled, code-rich summaries):

Try distilled references (above) first. Use upstream docs below only when more detail is needed.

Original upstream docs (for additional detail beyond the reference files):

Quick Reference

Connect (Classic Mode)

import rpyc
conn = rpyc.classic.connect("hostname")       # port 18812
conn.modules.os.listdir("/tmp")                # remote module access
conn.builtins.open("/etc/hostname").read()      # remote builtins

Create a Service

import rpyc
from rpyc.utils.server import ThreadedServer

class MyService(rpyc.Service):
    def on_connect(self, conn): pass
    def on_disconnect(self, conn): pass
    def exposed_add(self, a, b):
        return a + b

ThreadedServer(MyService, port=18861).start()

Connect to a Service

conn = rpyc.connect("hostname", 18861)
conn.root.add(3, 4)  # => 7

For async, timed calls, BgServingThread examples see async-and-events.md. For SSL, zero-deploy, service discovery examples see security-and-connections.md and services-and-servers.md.

Common Mistakes

Key Concepts

Transparent (remote objects behave local) | Symmetric (both ends serve requests) | Boxing (immutables by value, rest by reference as netrefs) | Capability-based security (pass specific objects, not broad access)