checkpoint-restore/criu

Features

• Application State Serialization - Converts a running application's memory, registers, and file descriptors into a persistent image for later reconstruction.
• Container Checkpointing - Freezes a running container process to disk to stop execution and save state for later resumption.
• Container Migration Engines - Functions as a container migration engine for transferring the live state of running containers between hosts.
• Process State Checkpointing - Saves running process states to disk and restarts them later from the exact point of suspension.
• Mount Namespace Dependency Resolution - Resolves mount namespace dependencies by specifying external bind mounts required to dump and restore a container.
• Checkpoint-Based Container Restoration - Resumes a container from a saved state by specifying an OCI image containing a checkpoint annotation.
• Container Restoration - Reloads a saved container state from disk to resume execution from the exact point of suspension.
• Cross-Container State Migration - Transfers a saved checkpoint from one container into a different container using a custom storage path.
• Live Container State Migration - Moves the live state of running containers across hosts or clusters to maintain service availability.
• Checkpoint Image Synthesis - Transforms a checkpoint archive into an annotated OCI image for distribution via a container registry.
• Network Connection Preservation - Captures and restores the state of TCP connections and Unix sockets to maintain connectivity after a restart.
• TCP Connection State Persisters - Saves and reconstructs network socket states to maintain active communication after a process restart.
• TCP State Reconstruction - Saves network socket queues and metadata to maintain active communication channels after a process is restored.
• Address Space Code Injection - Injects and runs binary blobs inside another running process to perform routines without disrupting the target task.
• Application File State Persistence - Captures and restores external bind mounts, mount devices, and file descriptors to maintain environment consistency.
• File Descriptor Restoration - Reconstitutes open files and service descriptors during recovery to ensure the application resumes from its exact state.
• Process and Memory Management - Freezes a running application and saves its memory and registers to disk for later resumption.
• Namespace-Based Isolation - Restores processes into dedicated PID and mount namespaces to prevent identifier conflicts and maintain environment isolation.
• PID Namespace Isolation - Creates a new PID namespace during restoration to prevent process ID conflicts when restarting applications.
• Live Process Code Injection - Inserts and executes binary blobs within live processes to extract internal state without termination.
• Mount Point Restoration - Saves and recreates filesystem mount points and namespaces to ensure the correct directory structure upon restoration.
• Process Checkpointing Tools - Provides the core ability to freeze running applications and save their memory and state to disk for later restoration.
• Process Identifier Restoration - Reconstructs process IDs and file descriptors to maintain the original relationship between parent and child processes.
• Pending Signal Preservation - Captures blocked signals and metadata during a freeze and reinstates them upon restart to maintain signal queues.
• Process State Managers - Captures and restores process trees, memory maps, and file descriptors within Linux namespaces.
• Process Tree Recovery - Returns a running process tree to a previously checkpointed state on the same machine.
• System Resource Serialization - Manages the serialization and recovery of Unix sockets, files, and network devices via plugin callbacks.
• TCP Connection Checkpointing - Saves the state of network sockets and their queues to enable the resumption of active communication after a process restart.
• TCP Connection Restoration - Reconstructs network sockets using saved state and queue data to resume communication from a previous point.
• Live Process Code Injection - Inserts and executes code within a live process to extract internal state information without terminating the application.
• State Snapshots - Saves a series of application states to disk to allow reverting to previous points for debugging.
• Diskless Replication - Transfers the state of a running application directly between hosts over a network without writing local files.
• Diskless Process Migration - Streams process images directly between hosts over a network to bypass local filesystem storage overhead.
• Checkpoint Distribution - Wraps container checkpoint data into standard OCI images for distribution and restoration through container registries.
• Incremental State Persistence - Saves only memory changes made since the last checkpoint to reduce storage requirements and backup time.
• Process State Duplication - Duplicates a process onto a remote machine periodically to ensure a quick failover if the primary system crashes.
• Checkpoint Image Conversion - Transforms process checkpoints into annotated OCI images for distribution via container registries.
• Process Network Restoration - Re-establishes network connectivity for a restored application, including the ability to change IP addresses across hosts.
• TCP Connection Repair - Offloads application-level request handling to another machine by repairing network connection state during a transition.
• Breakable Unix Socket Management - Marks local unix connections as breakable during a state dump so they can be re-established upon restoration.
• Boot Sequence Bypasses - Bypasses slow startup procedures by capturing the state of a service after its initial setup for faster subsequent launches.
• Checkpoint Operation Configuration - Sets parameters for dump and restore actions including process IDs, image directories, and network socket handling.
• External Network State Persistence - Handles the persistence and restoration of IP addresses, UNIX sockets, and virtual Ethernet devices.
• External State Integration - Incorporates data from third-party system tools such as network routing tables into the checkpoint image.
• Filesystem Lock Preservation - Captures held filesystem locks during a checkpoint to recreate them upon restoration of the process.
• GPU Workload Checkpointing - Captures and restores the state of applications using GPU acceleration to enable failure recovery and workload migration.
• Iterative Memory Migration - Implements iterative memory transfer to reduce the duration a process remains frozen during host migration.
• Memory Change Tracking - Monitors modifications to memory pages to optimize the dumping process and minimize the time an application remains frozen.
• Memory Image Deduplication - Merges identical memory pages across checkpoints to reduce the total storage space required for saved states.
• Post-Copy Memory Restoration - Implements a page-server mechanism to restore process execution immediately and fetch memory pages on demand.
• Non-Freezing Memory Capture - Reads process memory using system calls to prevent task freezes and memory exhaustion during the data dump phase.
• Process Restorability Verification - Checks a running process to determine if its current state can be successfully frozen and restored.
• Process Restoration Configuration - Uses configuration files to set behavioral flags such as preserving established TCP connections during a restore.
• Remote System Call Execution - Triggers system calls on remote targets to coordinate the transfer of process states.
• Seamless Kernel Upgrades - Checkpoints active processes before a kernel replacement and restores them afterward to maintain service continuity.
• JVM Instance Snapshotting - Captures the state of an executing Java program to eliminate start-up and warm-up delays during restoration.
• Breakable - Marks open network sockets as breakable during a state dump and reconnects them upon restoration.
• User-Space Probe Checkpointing - Captures the state of running applications using user-space probes without losing the probe memory mapping.
• Process Freeze Controls - Provides a background service to trigger application freezes and restores without direct command line interaction.
• Restore Image Validation - Checks the integrity and availability of required files before resuming a process to prevent crashes.

Star history