checkpoint-restore 开源项目推荐（28 个）

deeplethe /sandstorm

Fork() for AI agent microVMs. Spawn 100 children in ~100ms from a warm parent; BRANCH a live VM in ~150ms. KVM-isolated, snapshot CoW.

SJTU-IPADS /PhoenixOS

Fast OS-level support for GPU checkpoint and restore

ai-dynamo /snapshot

Snapshot gets GPU pods ready in seconds instead of minutes — by restoring a fully initialized GPU worker instead of starting one from scratch. Kubernetes-native checkpoint & restore that runs alongside your existing stack.

TensorRails /warmctl

GPU checkpoint/restore for vLLM: snapshot a warmed-up engine (compiled kernels, CUDA graphs, weights, 1-8 GPUs) into an OCI image and restore it in seconds instead of a multi-minute cold start. Scale to zero, spot GPUs, Kubernetes optional.

agent-substrate /always-on-agent

Run always-on messaging agents on Agent Substrate. A multi-tenant gateway holds the channel connections while each conversation runs as a suspendable actor, so presence stays warm and cognition suspends to reclaim compute. Reference implementation built on OpenClaw.

ammmir /sandboxer

Forkable code execution server for LLMs, agents, and devs

spacechunks /explorer

Host and explore Minecraft minigames

Alex-XJK /waypoint

Branchable agent containers: checkpoint a running sandbox (files, memory and live terminal sessions) and fork it into many independent live copies. Built on CRIU and OverlayFS.

DPicillo /paguro

Live migration of running Kubernetes pods between nodes – memory, TCP/UDP connections and volumes move while clients stay connected (CRIU + eBPF)

utnslab /Medes

Deduplication over dis-aggregated memory for Serverless Computing

seqeralabs /criu-static

Static build of CRIU (Checkpoint/Restore tool)

rugbedbugg /Resurrection

A userspace checkpoint/restore system for Linux processes. Picks restore per memory region with a caliberated cost model

mbien /JCRIU

JCRIU - Java binding for CRIU

udzura /ruby-criu

A libcriu wrapper for CRuby

Jordannst /mini-criu

Educational Linux checkpoint/restore project in C, focused on ptrace, /proc memory dumps, and experimental restore diagnostics.

haidinhtuan /shadow

SHADOW: Seamless Handoff And Zero-Downtime Orchestrated Workload Migration for Stateful Microservices — a Kubernetes operator for live migration with CRIU checkpointing and message replay

eddmpython /pyproc

Real Python runtime in the browser tab: processes and multicore parallelism, checkpoint/time-travel, portable .pymachine images, uv-grade instant environments, a Python server on a real URL. No server, no build step.

ratwolfzero /Hopalong_Map_Next_Gen

Two-pass Hopalong map: serial pass records segment checkpoints + bounds; parallel pass replays independent segments into private histograms. Fast, low-memory, bit-identical results.

VivianSobers /gpunap

Pause CUDA jobs to host RAM with NVIDIA's driver checkpoint API so a shared GPU box can be time-shared. Plan, feasibility probe, and a test suite recording what works and what breaks.

douvraslabs /chronos-cli

Chronos é uma proposta de plataforma de controle de estado para processos em execução, inspirada na ideia de transformar a memória RAM em um sistema de versionamento vivo.