feat(chrek): config refactor, /dev/shm support, and mount-policy rewrite (#5946)

components/src/dynamo/vllm/args.py

@@ -90,9 +90,6 @@ class Config:
     # Use vLLM's tokenizer for pre/post processing
     use_vllm_tokenizer: bool = False
 
-    # sleep mode support (enable_sleep_mode comes from vLLM's engine_args)
-    sleep_mode_level: int = 1
-
     # Whether to enable NATS for KV events (derived from kv_events_config in overwrite_args)
     use_kv_events: bool = False
 
@@ -301,13 +298,6 @@ def parse_args() -> Config:
         default=False,
         help="Use vLLM's tokenizer for pre and post processing. This bypasses Dynamo's preprocessor and only v1/chat/completions will be available through the Dynamo frontend.",
     )
-    parser.add_argument(
-        "--sleep-mode-level",
-        type=int,
-        default=1,
-        choices=[1, 2, 3],
-        help="Sleep mode level (1=offload to CPU, 2=discard weights, 3=discard all). Default: 1",
-    )
     add_config_dump_args(parser)
 
     parser = AsyncEngineArgs.add_cli_args(parser)
@@ -454,7 +444,6 @@ def parse_args() -> Config:
     config.enable_local_indexer = not args.durable_kv_events
     # For omni mode, use vLLM (AsyncOmni) tokenizer on backend
     config.use_vllm_tokenizer = args.use_vllm_tokenizer or args.omni
-    config.sleep_mode_level = args.sleep_mode_level
     # use_kv_events is set later in overwrite_args() based on kv_events_config
 
     # Validate custom Jinja template file exists if provided

components/src/dynamo/vllm/chrek.py

+# SPDX-FileCopyrightText: Copyright (c) 2025-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+"""
+Checkpoint/restore (chrek) integration for vLLM workers.
+
+Handles the checkpoint job pod lifecycle:
+1. Early exit if a checkpoint already exists (idempotency)
+2. Sleep model for CRIU-friendly GPU state
+3. Signal readiness for DaemonSet to begin checkpoint
+4. Poll for checkpoint completion or CRIU restore detection
+5. Wake model after restore
+
+Environment variables (all required in checkpoint mode, no fallbacks):
+- DYN_CHECKPOINT_SIGNAL_FILE: Path where DaemonSet writes completion signal
+- DYN_READY_FOR_CHECKPOINT_FILE: Path where this worker writes readiness marker
+- DYN_CHECKPOINT_STORAGE_TYPE: Storage backend (pvc, s3, oci)
+- DYN_CHECKPOINT_LOCATION: Full checkpoint path (for idempotency check)
+- DYN_RESTORE_MARKER_FILE: Path written by restore-entrypoint before CRIU restore
+"""
+
+import asyncio
+import json
+import logging
+import os
+from typing import Optional
+
+logger = logging.getLogger(__name__)
+
+_REQUIRED_ENV_VARS = [
+    "DYN_CHECKPOINT_SIGNAL_FILE",
+    "DYN_READY_FOR_CHECKPOINT_FILE",
+    "DYN_CHECKPOINT_STORAGE_TYPE",
+    "DYN_CHECKPOINT_LOCATION",
+    "DYN_RESTORE_MARKER_FILE",
+]
+
+
+class CheckpointConfig:
+    """Parsed and validated checkpoint configuration from environment variables."""
+
+    def __init__(self):
+        self.signal_file = os.environ["DYN_CHECKPOINT_SIGNAL_FILE"]
+        self.ready_file = os.environ["DYN_READY_FOR_CHECKPOINT_FILE"]
+        self.storage_type = os.environ["DYN_CHECKPOINT_STORAGE_TYPE"]
+        self.location = os.environ["DYN_CHECKPOINT_LOCATION"]
+        self.restore_marker = os.environ["DYN_RESTORE_MARKER_FILE"]
+
+    def _read_status_file(self, path: str) -> dict:
+        with open(path) as f:
+            status = json.load(f)
+
+        success = status.get("success")
+        if not isinstance(success, bool):
+            raise ValueError(f"missing or invalid success field in {path}")
+        return status
+
+    def checkpoint_exists(self) -> bool:
+        """Check if a completed checkpoint already exists (idempotency).
+
+        For PVC storage, checks for checkpoint.done marker at the location.
+        Returns True if the job should exit without loading the model.
+        """
+        assert (
+            self.storage_type == "pvc"
+        ), "Checkpoint existence check is only implemented for PVC storage"
+        if self.storage_type == "pvc" and self.location:
+            done_marker = f"{self.location}/checkpoint.done"
+            if os.path.exists(done_marker):
+                try:
+                    status = self._read_status_file(done_marker)
+                except (OSError, ValueError, json.JSONDecodeError) as exc:
+                    logger.warning(
+                        f"Invalid checkpoint.done marker at {done_marker}, ignoring stale checkpoint: {exc}"
+                    )
+                    return False
+
+                if status["success"]:
+                    logger.info(
+                        f"Existing successful checkpoint found at {self.location}, skipping"
+                    )
+                    return True
+
+                logger.warning(
+                    f"Existing checkpoint marker reports failure at {self.location}: "
+                    f"{status.get('error', 'unknown error')}"
+                )
+                return False
+
+            logger.info(f"No checkpoint at {self.location}, creating new one")
+        return False
+
+    async def run_lifecycle(self, engine_client, sleep_level: int) -> bool:
+        """Run the full checkpoint lifecycle after the engine is loaded.
+
+        1. Put model to sleep (CRIU-friendly GPU state)
+        2. Write ready file (triggers DaemonSet checkpoint via readiness probe)
+        3. Poll for signal file (checkpoint done) or restore marker (CRIU restored us)
+        4. If restored: wake model and return True (caller proceeds with registration)
+        5. If checkpoint done: return False (caller should exit)
+        """
+        # Sleep model for checkpoint
+        logger.info(f"Putting model to sleep (level={sleep_level})")
+        await engine_client.sleep(level=sleep_level)
+
+        # Signal readiness
+        with open(self.ready_file, "w") as f:
+            f.write("ready")
+        logger.info(
+            f"Ready for checkpoint. Waiting for signal: {self.signal_file} "
+            f"or restore marker: {self.restore_marker}"
+        )
+
+        # Poll for signal or restore
+        while True:
+            if os.path.exists(self.restore_marker):
+                logger.info(f"Restore detected (marker: {self.restore_marker})")
+                logger.info("Waking up model after restore")
+                await engine_client.wake_up()
+                return True
+
+            if os.path.exists(self.signal_file):
+                try:
+                    signal = self._read_status_file(self.signal_file)
+                except (OSError, ValueError, json.JSONDecodeError) as exc:
+                    raise RuntimeError(
+                        f"Invalid checkpoint signal file {self.signal_file}: {exc}"
+                    ) from exc
+
+                if signal["success"]:
+                    logger.info(f"Checkpoint complete (signal: {self.signal_file})")
+                    return False
+
+                raise RuntimeError(
+                    f"Checkpoint failed (signal: {self.signal_file}): "
+                    f"{signal.get('error', 'unknown error')}"
+                )
+
+            await asyncio.sleep(1)
+
+
+def get_checkpoint_config() -> Optional[CheckpointConfig]:
+    """Returns CheckpointConfig if in checkpoint mode, None otherwise.
+
+    Checkpoint mode is detected by DYN_CHECKPOINT_SIGNAL_FILE being set.
+    If in checkpoint mode, all required env vars must be present — raises
+    EnvironmentError if any are missing.
+    """
+    if "DYN_CHECKPOINT_SIGNAL_FILE" not in os.environ:
+        return None
+
+    missing = [v for v in _REQUIRED_ENV_VARS if v not in os.environ]
+    if missing:
+        raise EnvironmentError(
+            f"Checkpoint mode requires these environment variables: {', '.join(missing)}"
+        )
+    return CheckpointConfig()

components/src/dynamo/vllm/main.py

@@ -56,6 +56,7 @@ from dynamo.vllm.multimodal_handlers import (
 from dynamo.vllm.multimodal_utils.encode_utils import create_ec_transfer_config
 
 from .args import Config, overwrite_args, parse_args
+from .chrek import get_checkpoint_config
 from .handlers import DecodeWorkerHandler, PrefillWorkerHandler
 from .health_check import (
     VllmHealthCheckPayload,
@@ -66,6 +67,7 @@ from .publisher import DYNAMO_COMPONENT_REGISTRY, StatLoggerFactory
 
 configure_dynamo_logging()
 logger = logging.getLogger(__name__)
+CHECKPOINT_SLEEP_MODE_LEVEL = 1
 
 
 async def _handle_non_leader_node(dp_rank: int) -> None:
@@ -81,46 +83,17 @@ async def _handle_non_leader_node(dp_rank: int) -> None:
     await asyncio.Event().wait()
 
 
-async def await_checkpoint_and_was_restored(signal_file: str) -> bool:
+async def graceful_shutdown(runtime, shutdown_event):
     """
-    Wait for checkpoint signal file OR restore marker file.
-
-    In checkpoint creation mode, poll until either:
-    1. The signal file exists (checkpoint complete, should exit)
-    2. The restore marker file exists (restored by CRIU, should proceed)
-
-    The restore marker file is created by the restore-entrypoint before CRIU restore,
-    so the restored process can detect it was restored even though os.environ is
-    restored from the checkpoint and doesn't contain new container env vars.
-
-    Args:
-        signal_file: Path to the checkpoint signal file
-
-    Returns:
-        True if restored (should proceed with registration)
-        False if signal file detected (should exit)
+    Shutdown dynamo distributed runtime.
+    The endpoints will be immediately invalidated so no new requests will be accepted.
+    For endpoints served with graceful_shutdown=True, the serving function will wait until all in-flight requests are finished.
+    For endpoints served with graceful_shutdown=False, the serving function will return immediately.
     """
-    # Get restore marker file path (created by restore entrypoint before CRIU restore)
-    restore_marker = os.environ.get("DYN_RESTORE_MARKER_FILE", "/tmp/dynamo-restored")
-
-    logger.info(
-        f"CHECKPOINT_READY: Model loaded, ready for container checkpoint. Waiting for signal file: {signal_file} or restore marker file: {restore_marker}"
-    )
-
-    while True:
-        # Check if we've been restored (marker file created by restore entrypoint)
-        if os.path.exists(restore_marker):
-            logger.info(
-                f"Detected restore from checkpoint (marker file exists: {restore_marker})"
-            )
-            return True  # Restored - proceed with registration
-
-        # Check if checkpoint is complete (signal file exists)
-        if os.path.exists(signal_file):
-            logger.info(f"Checkpoint signal file detected: {signal_file}")
-            return False  # Checkpoint done - exit
-
-        await asyncio.sleep(1)
+    logging.info("Received shutdown signal, shutting down DistributedRuntime")
+    shutdown_event.set()
+    runtime.shutdown()
+    logging.info("DistributedRuntime shutdown complete")
 
 
 async def worker():
@@ -134,29 +107,10 @@ async def worker():
     if not config.served_model_name:
         config.served_model_name = config.engine_args.served_model_name = config.model
 
-    # Check checkpoint-related environment variables EARLY
-    signal_file = os.environ.get("DYN_CHECKPOINT_SIGNAL_FILE")
-    ready_file = os.environ.get("DYN_CHECKPOINT_READY_FILE")
-
-    is_checkpoint_mode = signal_file is not None
-
-    # EARLY EXIT: Check if checkpoint already exists (before downloading model!)
-    if is_checkpoint_mode:
-        storage_type = os.environ.get("DYN_CHECKPOINT_STORAGE_TYPE")
-        checkpoint_location = os.environ.get("DYN_CHECKPOINT_LOCATION")
-
-        if storage_type == "pvc" and checkpoint_location:
-            done_marker = f"{checkpoint_location}/checkpoint.done"
-
-            if os.path.exists(done_marker):
-                logger.info(
-                    f"Found existing checkpoint at {checkpoint_location}. Storage type: {storage_type}"
-                )
-                return
-            else:
-                logger.info(
-                    f"Checkpoint not found at: {checkpoint_location}. creating new checkpoint"
-                )
+    # Check checkpoint mode and validate env vars EARLY (fail fast if misconfigured)
+    checkpoint_cfg = get_checkpoint_config()
+    if checkpoint_cfg and checkpoint_cfg.checkpoint_exists():
+        return
 
     # Download the model if necessary using modelexpress.
     # We want it on disk before we start vllm to avoid downloading from HuggingFace.
@@ -173,39 +127,20 @@ async def worker():
     # CHECKPOINT MODE: Load engine BEFORE runtime creation
     # This allows checkpointing GPU state before runtime connections are established
     pre_created_engine = None
-    is_restored = False
-    if is_checkpoint_mode:
+    if checkpoint_cfg is not None:
         logger.info(
-            f"Checkpoint mode enabled (DYN_CHECKPOINT_SIGNAL_FILE={signal_file})"
+            f"Checkpoint mode enabled (signal_file={checkpoint_cfg.signal_file})"
         )
 
-        # CHECKPOINT MODE: Load model, sleep, wait for signal file or restore
+        # Checkpoint mode requires sleep mode — enable before engine init
+        config.engine_args.enable_sleep_mode = True
+
         pre_created_engine = setup_vllm_engine(config)
         engine_client = pre_created_engine[0]
 
-        # Put model to sleep before checkpoint (if sleep mode enabled)
-        if config.engine_args.enable_sleep_mode:
-            logger.info(f"Putting model to sleep (level={config.sleep_mode_level})")
-            await engine_client.sleep(level=config.sleep_mode_level)
-
-        # Write ready file to signal that we're ready for checkpointing
-        if ready_file:
-            with open(ready_file, "w") as f:
-                f.write("ready")
-            logger.info(f"Wrote checkpoint ready file: {ready_file}")
-
-        # Wait for checkpoint signal file OR restore detection
-        is_restored = await await_checkpoint_and_was_restored(signal_file)
-
-        if is_restored:
-            # Wake up model and proceed with registration
-            if config.engine_args.enable_sleep_mode:
-                logger.info("Waking up model after checkpoint restore")
-                await engine_client.wake_up()
-            logger.info("Proceeding with endpoint registration after restore")
-        else:
-            # Checkpoint complete, exit
-            logger.info("Exiting after checkpoint completion")
+        if not await checkpoint_cfg.run_lifecycle(
+            engine_client, CHECKPOINT_SLEEP_MODE_LEVEL
+        ):
             return
 
     shutdown_event = asyncio.Event()

deploy/chrek/Dockerfile

@@ -142,14 +142,6 @@ COPY --from=builder /restore-entrypoint /restore-entrypoint
 # Create checkpoint directory
 RUN mkdir -p /checkpoints
 
-# Set environment variables
-ENV HOST_PROC=/host/proc \
-    CONTAINERD_SOCKET=/run/containerd/containerd.sock \
-    CHECKPOINT_DIR=/checkpoints \
-    LISTEN_ADDR=:8080
-
-EXPOSE 8080
-
 USER root
 
 ENTRYPOINT ["/usr/local/bin/chrek-agent"]
@@ -172,7 +164,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     libnl-3-200 \
     libnl-route-3-200 \
     libprotobuf-c1 \
-    libgnutls30 \
+    libgnutls30t64 \
     libnftables1 \
     iproute2 \
     iptables \
@@ -191,23 +183,11 @@ COPY --from=criu-builder /tmp/cuda-checkpoint/bin/x86_64_Linux/cuda-checkpoint /
 RUN chmod +x /usr/local/sbin/cuda-checkpoint
 
 # Create directories
-RUN mkdir -p /checkpoints /var/run/criu /tmp /var/criu-work
+RUN mkdir -p /checkpoints /var/run/criu /var/criu-work
 
 # Copy restore binaries
 COPY --from=builder /restore-entrypoint /restore-entrypoint
 RUN chmod +x /restore-entrypoint
 
-COPY scripts/smart-entrypoint.sh /smart-entrypoint.sh
-RUN chmod +x /smart-entrypoint.sh
-
-# Set environment variables
-ENV DYN_CHECKPOINT_PATH=/checkpoints \
-    RESTORE_TRIGGER=/tmp/restore-trigger \
-    RESTORE_WAIT_TIMEOUT=300 \
-    CRIU_LOG_LEVEL=4 \
-    WAIT_FOR_CHECKPOINT=0 \
-    CUDA_PLUGIN_DIR=/usr/local/lib/criu \
-    DEBUG=0
-
-ENTRYPOINT ["/smart-entrypoint.sh"]
+ENTRYPOINT ["/restore-entrypoint"]
 CMD []

deploy/chrek/cmd/agent/config.go

+// config.go provides configuration loading for the checkpoint agent.
+package main
+
+import (
+	"fmt"
+	"os"
+
+	"gopkg.in/yaml.v3"
+
+	"github.com/ai-dynamo/dynamo/deploy/chrek/pkg/checkpoint"
+)
+
+// ConfigMapPath is the default path where the ConfigMap is mounted.
+const ConfigMapPath = "/etc/chrek/config.yaml"
+
+// CheckpointSignalSource determines how checkpoint operations are triggered.
+type CheckpointSignalSource string
+
+const (
+	// SignalFromHTTP triggers checkpoints via HTTP API requests.
+	SignalFromHTTP CheckpointSignalSource = "http"
+	// SignalFromWatcher triggers checkpoints automatically when pods become Ready.
+	SignalFromWatcher CheckpointSignalSource = "watcher"
+)
+
+// FullConfig is the root configuration structure loaded from the ConfigMap.
+type FullConfig struct {
+	Agent      AgentConfig               `yaml:"agent"`
+	Checkpoint checkpoint.CheckpointSpec `yaml:"checkpoint"`
+}
+
+// AgentConfig holds the runtime configuration for the checkpoint agent daemon.
+type AgentConfig struct {
+	// SignalSource determines how checkpoints are triggered: "http" or "watcher"
+	SignalSource string `yaml:"signalSource"`
+
+	// ListenAddr is the HTTP server address for health checks and API
+	ListenAddr string `yaml:"listenAddr"`
+
+	// NodeName is the Kubernetes node name (from NODE_NAME env, downward API)
+	NodeName string `yaml:"-"`
+
+	// RestrictedNamespace restricts pod watching to this namespace (optional)
+	RestrictedNamespace string `yaml:"-"`
+}
+
+// ConfigError represents a configuration validation error.
+type ConfigError struct {
+	Field   string
+	Message string
+}
+
+func (e *ConfigError) Error() string {
+	return fmt.Sprintf("config error: %s: %s", e.Field, e.Message)
+}
+
+// LoadConfig loads the full configuration from a YAML file.
+func LoadConfig(path string) (*FullConfig, error) {
+	data, err := os.ReadFile(path)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read config file %s: %w", path, err)
+	}
+
+	cfg := &FullConfig{}
+	if err := yaml.Unmarshal(data, cfg); err != nil {
+		return nil, fmt.Errorf("failed to parse config file %s: %w", path, err)
+	}
+
+	// Apply environment variable overrides
+	cfg.Agent.loadEnvOverrides()
+
+	return cfg, nil
+}
+
+// LoadConfigOrDefault loads configuration from a file, falling back to zero values if the file doesn't exist.
+func LoadConfigOrDefault(path string) (*FullConfig, error) {
+	cfg, err := LoadConfig(path)
+	if err != nil {
+		if os.IsNotExist(err) {
+			cfg = &FullConfig{}
+			cfg.Agent.loadEnvOverrides()
+			return cfg, nil
+		}
+		return nil, err
+	}
+	return cfg, nil
+}
+
+// loadEnvOverrides applies environment variable overrides to the AgentConfig.
+func (c *AgentConfig) loadEnvOverrides() {
+	if v := os.Getenv("NODE_NAME"); v != "" {
+		c.NodeName = v
+	}
+	if v := os.Getenv("RESTRICTED_NAMESPACE"); v != "" {
+		c.RestrictedNamespace = v
+	}
+}
+
+// GetSignalSource returns the signal source as a CheckpointSignalSource type.
+func (c *AgentConfig) GetSignalSource() CheckpointSignalSource {
+	return CheckpointSignalSource(c.SignalSource)
+}
+
+// Validate checks that the AgentConfig has valid values.
+func (c *AgentConfig) Validate() error {
+	if c.SignalSource != string(SignalFromHTTP) && c.SignalSource != string(SignalFromWatcher) {
+		return &ConfigError{
+			Field:   "signalSource",
+			Message: "must be 'http' or 'watcher'",
+		}
+	}
+	if c.SignalSource == string(SignalFromHTTP) && c.ListenAddr == "" {
+		return &ConfigError{
+			Field:   "listenAddr",
+			Message: "cannot be empty when signalSource is 'http'",
+		}
+	}
+	return nil
+}
+
+// Validate validates the full configuration.
+func (c *FullConfig) Validate() error {
+	if err := c.Agent.Validate(); err != nil {
+		return err
+	}
+	if err := c.Checkpoint.Validate(); err != nil {
+		return err
+	}
+	return nil
+}

deploy/chrek/cmd/agent/main.go

@@ -6,174 +6,39 @@ package main
 
 import (
 	"context"
-	"encoding/json"
-	"fmt"
 	"log"
 	"net/http"
 	"os"
 	"os/signal"
-	"strconv"
-	"strings"
 	"syscall"
 	"time"
 
 	"github.com/ai-dynamo/dynamo/deploy/chrek/pkg/checkpoint"
-	checkpointk8s "github.com/ai-dynamo/dynamo/deploy/chrek/pkg/checkpoint/k8s"
-	"github.com/ai-dynamo/dynamo/deploy/chrek/pkg/common"
+	httpApiServer "github.com/ai-dynamo/dynamo/deploy/chrek/pkg/http_api_server"
 	"github.com/ai-dynamo/dynamo/deploy/chrek/pkg/watcher"
 )
 
-// CheckpointSignalSource determines how checkpoint operations are triggered
-type CheckpointSignalSource string
-
-const (
-	// SignalFromHTTP triggers checkpoints via HTTP API requests
-	SignalFromHTTP CheckpointSignalSource = "http"
-	// SignalFromWatcher triggers checkpoints automatically when pods become Ready
-	SignalFromWatcher CheckpointSignalSource = "watcher"
-)
-
-// Config holds the agent configuration
-type Config struct {
-	// Common settings
-	ContainerdSocket    string
-	CheckpointDir       string
-	HostProc            string
-	NodeName            string
-	RestrictedNamespace string // Optional: restrict pod watching to this namespace
-
-	// Mode selection
-	SignalSource CheckpointSignalSource // "http" or "watcher"
-
-	// HTTP API mode settings (used when SignalSource = "http")
-	ListenAddr string
-
-	// CRIU settings (configurable options only; LeaveRunning, ShellJob, etc. are hardcoded in pkg/checkpoint/criu.go)
-	CUDAPluginDir  string   // Path to CRIU CUDA plugin directory
-	GhostLimit     uint32   // CRIU ghost limit in bytes
-	Timeout        uint32   // CRIU timeout in seconds
-	ExternalMounts []string // External mount mappings
-}
-
-// Server is the HTTP API server
-type Server struct {
-	config          Config
-	discoveryClient *checkpointk8s.DiscoveryClient
-	checkpointer    *checkpoint.Checkpointer
-}
-
-// CheckpointRequest is the request body for checkpoint operations
-type CheckpointRequest struct {
-	ContainerID  string `json:"container_id"`
-	CheckpointID string `json:"checkpoint_id"`
-	PodName      string `json:"pod_name,omitempty"`
-	PodNamespace string `json:"pod_namespace,omitempty"`
-	DisableCUDA  bool   `json:"disable_cuda,omitempty"` // Disable CUDA plugin for non-GPU workloads
-}
-
-// TriggerRestoreRequest is the request body for Option A self-restoring trigger
-type TriggerRestoreRequest struct {
-	CheckpointID           string `json:"checkpoint_id"`
-	PlaceholderContainerID string `json:"placeholder_container_id"`
-	SkipImageValidation    bool   `json:"skip_image_validation,omitempty"` // Skip image matching check
-}
-
-// TriggerRestoreResponse is the response for trigger restore operations
-type TriggerRestoreResponse struct {
-	Success          bool   `json:"success"`
-	Message          string `json:"message,omitempty"`
-	Error            string `json:"error,omitempty"`
-	TriggerPath      string `json:"trigger_path,omitempty"`
-	CheckpointImage  string `json:"checkpoint_image,omitempty"`
-	PlaceholderImage string `json:"placeholder_image,omitempty"`
-}
-
-// CheckpointResponse is the response for checkpoint operations
-type CheckpointResponse struct {
-	Success      bool   `json:"success"`
-	CheckpointID string `json:"checkpoint_id,omitempty"`
-	Message      string `json:"message,omitempty"`
-	Error        string `json:"error,omitempty"`
-}
-
-// CheckpointInfo represents information about a checkpoint
-type CheckpointInfo struct {
-	ID           string    `json:"id"`
-	CreatedAt    time.Time `json:"created_at"`
-	SourceNode   string    `json:"source_node"`
-	ContainerID  string    `json:"container_id"`
-	PodName      string    `json:"pod_name"`
-	PodNamespace string    `json:"pod_namespace"`
-	Image        string    `json:"image"`
-}
-
-// ListCheckpointsResponse is the response for list checkpoints
-type ListCheckpointsResponse struct {
-	Checkpoints []CheckpointInfo `json:"checkpoints"`
-}
-
-// HealthResponse is the response for health check
-type HealthResponse struct {
-	Status   string `json:"status"`
-	NodeName string `json:"node_name"`
-}
-
 func main() {
-	// Parse signal source - default to HTTP for backward compatibility
-	signalSource := CheckpointSignalSource(strings.ToLower(getEnv("CHECKPOINT_SIGNAL_FROM", "http")))
-	if signalSource != SignalFromHTTP && signalSource != SignalFromWatcher {
-		log.Fatalf("Invalid CHECKPOINT_SIGNAL_FROM value: %q (must be 'http' or 'watcher')", signalSource)
-	}
-
-	// Parse CRIU settings
-	var ghostLimit, timeout uint32
-	if v := os.Getenv("CRIU_GHOST_LIMIT"); v != "" {
-		if parsed, err := strconv.ParseUint(v, 10, 32); err == nil {
-			ghostLimit = uint32(parsed)
-		}
-	}
-	if v := os.Getenv("CRIU_TIMEOUT"); v != "" {
-		if parsed, err := strconv.ParseUint(v, 10, 32); err == nil {
-			timeout = uint32(parsed)
-		}
-	}
-
-	// Parse external mounts (comma-separated)
-	var externalMounts []string
-	if v := os.Getenv("EXTERNAL_MOUNTS"); v != "" {
-		externalMounts = strings.Split(v, ",")
+	// Load configuration from ConfigMap (or use defaults if not found)
+	cfg, err := LoadConfigOrDefault(ConfigMapPath)
+	if err != nil {
+		log.Fatalf("Failed to load configuration: %v", err)
 	}
 
-	config := Config{
-		// Common settings
-		ContainerdSocket:    getEnv("CONTAINERD_SOCKET", "/run/containerd/containerd.sock"),
-		CheckpointDir:       getEnv("CHECKPOINT_DIR", "/checkpoints"),
-		HostProc:            getEnv("HOST_PROC", "/host/proc"),
-		NodeName:            getEnv("NODE_NAME", "unknown"),
-		RestrictedNamespace: os.Getenv("RESTRICTED_NAMESPACE"), // Optional: empty = cluster-wide watching
-
-		// Mode selection
-		SignalSource: signalSource,
-
-		// HTTP API settings
-		ListenAddr: getEnv("LISTEN_ADDR", ":8080"),
-
-		// CRIU settings
-		CUDAPluginDir:  getEnv("CUDA_PLUGIN_DIR", ""),
-		GhostLimit:     ghostLimit,
-		Timeout:        timeout,
-		ExternalMounts: externalMounts,
+	// Validate configuration
+	if err := cfg.Agent.Validate(); err != nil {
+		log.Fatalf("Invalid configuration: %v", err)
 	}
 
 	// Create discovery client
-	discoveryClient, err := checkpointk8s.NewDiscoveryClient(config.ContainerdSocket)
+	discoveryClient, err := checkpoint.NewDiscoveryClient()
 	if err != nil {
 		log.Fatalf("Failed to create discovery client: %v", err)
 	}
 	defer discoveryClient.Close()
 
 	// Create checkpointer
-	checkpointer := checkpoint.NewCheckpointer(discoveryClient, config.HostProc)
+	checkpointer := checkpoint.NewCheckpointer(discoveryClient)
 
 	// Context for graceful shutdown
 	ctx, cancel := context.WithCancel(context.Background())
@@ -183,60 +48,39 @@ func main() {
 	sigChan := make(chan os.Signal, 1)
 	signal.Notify(sigChan, syscall.SIGINT, syscall.SIGTERM)
 
-	log.Printf("CRIU Node Agent starting (node: %s)", config.NodeName)
-	log.Printf("Checkpoint directory: %s", config.CheckpointDir)
-	log.Printf("Signal source: %s", config.SignalSource)
+	log.Printf("CRIU Node Agent starting (node: %s)", cfg.Agent.NodeName)
+	log.Printf("Checkpoint directory: %s", cfg.Checkpoint.BasePath)
+	log.Printf("Signal source: %s", cfg.Agent.SignalSource)
 
-	switch config.SignalSource {
+	switch cfg.Agent.GetSignalSource() {
 	case SignalFromHTTP:
-		server := &Server{
-			config:          config,
-			discoveryClient: discoveryClient,
-			checkpointer:    checkpointer,
-		}
-
-		// Setup routes
-		mux := http.NewServeMux()
-		mux.HandleFunc("/health", server.handleHealth)
-		mux.HandleFunc("/checkpoint", server.handleCheckpoint)
-		mux.HandleFunc("/restore/trigger", server.handleTriggerRestore)
-		mux.HandleFunc("/checkpoints", server.handleListCheckpoints)
-
-		httpServer := &http.Server{
-			Addr:         config.ListenAddr,
-			Handler:      loggingMiddleware(mux),
-			ReadTimeout:  30 * time.Second,
-			WriteTimeout: 300 * time.Second,
-			IdleTimeout:  120 * time.Second,
+		serverCfg := httpApiServer.ServerConfig{
+			ListenAddr:     cfg.Agent.ListenAddr,
+			NodeName:       cfg.Agent.NodeName,
+			CheckpointSpec: &cfg.Checkpoint,
 		}
+		srv := httpApiServer.NewServer(serverCfg, checkpointer)
 
 		// Handle graceful shutdown
 		go func() {
 			<-sigChan
-			log.Println("Shutting down HTTP server...")
 			shutdownCtx, shutdownCancel := context.WithTimeout(context.Background(), 30*time.Second)
 			defer shutdownCancel()
-			if err := httpServer.Shutdown(shutdownCtx); err != nil {
+			if err := srv.Shutdown(shutdownCtx); err != nil {
 				log.Printf("HTTP server shutdown error: %v", err)
 			}
 		}()
 
-		log.Printf("HTTP API server listening on %s", config.ListenAddr)
-		if err := httpServer.ListenAndServe(); err != http.ErrServerClosed {
+		if err := srv.Start(); err != http.ErrServerClosed {
 			log.Fatalf("HTTP server error: %v", err)
 		}
 
 	case SignalFromWatcher:
-		watcherConfig := watcher.Config{
-			NodeName:            config.NodeName,
-			CheckpointDir:       config.CheckpointDir,
-			HostProc:            config.HostProc,
-			ListenAddr:          config.ListenAddr, // For health check endpoint
-			RestrictedNamespace: config.RestrictedNamespace,
-			CUDAPluginDir:       config.CUDAPluginDir,
-			GhostLimit:          config.GhostLimit,
-			Timeout:             config.Timeout,
-			ExternalMounts:      config.ExternalMounts,
+		watcherConfig := watcher.WatcherConfig{
+			NodeName:            cfg.Agent.NodeName,
+			ListenAddr:          cfg.Agent.ListenAddr,
+			RestrictedNamespace: cfg.Agent.RestrictedNamespace,
+			CheckpointSpec:      &cfg.Checkpoint,
 		}
 
 		podWatcher, err := watcher.NewWatcher(watcherConfig, discoveryClient, checkpointer)
@@ -251,304 +95,15 @@ func main() {
 			cancel()
 		}()
 
-		log.Printf("Pod watcher started (watching for label: nvidia.com/checkpoint-source=true)")
-		log.Printf("Health check endpoint: http://0.0.0.0%s/health", config.ListenAddr)
+		log.Printf("Pod watcher started (watching for label: %s=true)", checkpoint.KubeLabelCheckpointSource)
+		log.Printf("Health check endpoint: http://0.0.0.0%s/health", cfg.Agent.ListenAddr)
 		if err := podWatcher.Start(ctx); err != nil {
 			log.Printf("Pod watcher error: %v", err)
 		}
-	}
-
-	log.Println("Agent stopped")
-}
-
-func (s *Server) handleHealth(w http.ResponseWriter, r *http.Request) {
-	if r.Method != http.MethodGet {
-		http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
-		return
-	}
-
-	resp := HealthResponse{
-		Status:   "healthy",
-		NodeName: s.config.NodeName,
-	}
 
-	writeJSON(w, http.StatusOK, resp)
-}
-
-func (s *Server) handleCheckpoint(w http.ResponseWriter, r *http.Request) {
-	if r.Method != http.MethodPost {
-		http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
-		return
-	}
-
-	var req CheckpointRequest
-	if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
-		writeJSON(w, http.StatusBadRequest, CheckpointResponse{
-			Success: false,
-			Error:   fmt.Sprintf("Invalid request body: %v", err),
-		})
-		return
-	}
-
-	if req.ContainerID == "" {
-		writeJSON(w, http.StatusBadRequest, CheckpointResponse{
-			Success: false,
-			Error:   "container_id is required",
-		})
-		return
-	}
-
-	if req.CheckpointID == "" {
-		req.CheckpointID = fmt.Sprintf("ckpt-%d", time.Now().UnixNano())
-	}
-
-	// Determine CUDA plugin directory - only use if not explicitly disabled
-	cudaPluginDir := s.config.CUDAPluginDir
-	if req.DisableCUDA {
-		cudaPluginDir = ""
-	}
-
-	// Parse optional CRIU settings from environment
-	var ghostLimit, timeout uint32
-	if v := os.Getenv("CRIU_GHOST_LIMIT"); v != "" {
-		if parsed, err := strconv.ParseUint(v, 10, 32); err == nil {
-			ghostLimit = uint32(parsed)
-		}
-	}
-	if v := os.Getenv("CRIU_TIMEOUT"); v != "" {
-		if parsed, err := strconv.ParseUint(v, 10, 32); err == nil {
-			timeout = uint32(parsed)
-		}
+	default:
+		log.Fatalf("Unknown signal source: %s", cfg.Agent.SignalSource)
 	}
 
-	opts := checkpoint.Options{
-		ContainerID:   req.ContainerID,
-		CheckpointID:  req.CheckpointID,
-		CheckpointDir: s.config.CheckpointDir,
-		NodeName:      s.config.NodeName,
-		PodName:       req.PodName,
-		PodNamespace:  req.PodNamespace,
-		GhostLimit:    ghostLimit,
-		Timeout:       timeout,
-		CUDAPluginDir: cudaPluginDir,
-	}
-
-	ctx := r.Context()
-	result, err := s.checkpointer.Checkpoint(ctx, opts)
-	if err != nil {
-		log.Printf("Checkpoint failed: %v", err)
-		writeJSON(w, http.StatusInternalServerError, CheckpointResponse{
-			Success: false,
-			Error:   err.Error(),
-		})
-		return
-	}
-
-	log.Printf("Checkpoint successful: %s", result.CheckpointID)
-	writeJSON(w, http.StatusOK, CheckpointResponse{
-		Success:      true,
-		CheckpointID: result.CheckpointID,
-		Message:      fmt.Sprintf("Checkpoint created successfully at %s", result.CheckpointDir),
-	})
-}
-
-// handleTriggerRestore implements Option A from RESTORE_ANALYSIS.md
-// It triggers a self-restoring placeholder container to start CRIU restore.
-// The agent writes a trigger file to the placeholder's filesystem, which
-// the placeholder's entrypoint script detects and uses to start restoration.
-func (s *Server) handleTriggerRestore(w http.ResponseWriter, r *http.Request) {
-	if r.Method != http.MethodPost {
-		http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
-		return
-	}
-
-	var req TriggerRestoreRequest
-	if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
-		writeJSON(w, http.StatusBadRequest, TriggerRestoreResponse{
-			Success: false,
-			Error:   fmt.Sprintf("Invalid request body: %v", err),
-		})
-		return
-	}
-
-	if req.CheckpointID == "" {
-		writeJSON(w, http.StatusBadRequest, TriggerRestoreResponse{
-			Success: false,
-			Error:   "checkpoint_id is required",
-		})
-		return
-	}
-
-	if req.PlaceholderContainerID == "" {
-		writeJSON(w, http.StatusBadRequest, TriggerRestoreResponse{
-			Success: false,
-			Error:   "placeholder_container_id is required",
-		})
-		return
-	}
-
-	// Verify checkpoint exists and load metadata
-	checkpointPath := common.GetCheckpointDir(s.config.CheckpointDir, req.CheckpointID)
-	checkpointMeta, err := common.LoadMetadata(checkpointPath)
-	if err != nil {
-		writeJSON(w, http.StatusNotFound, TriggerRestoreResponse{
-			Success: false,
-			Error:   fmt.Sprintf("Checkpoint not found: %v", err),
-		})
-		return
-	}
-
-	// Resolve placeholder container to get PID and image
-	ctx := r.Context()
-	containerInfo, err := s.discoveryClient.ResolveContainer(ctx, req.PlaceholderContainerID)
-	if err != nil {
-		writeJSON(w, http.StatusInternalServerError, TriggerRestoreResponse{
-			Success: false,
-			Error:   fmt.Sprintf("Failed to resolve placeholder container: %v", err),
-		})
-		return
-	}
-
-	// Validate that placeholder image matches checkpoint's original image
-	// This is critical because rootfs-diff.tar only contains upperdir modifications,
-	// not the base image layers (lowerdir). If images differ, CRIU restore will fail.
-	if !req.SkipImageValidation && checkpointMeta.Image != "" {
-		if !imagesCompatible(checkpointMeta.Image, containerInfo.Image) {
-			writeJSON(w, http.StatusBadRequest, TriggerRestoreResponse{
-				Success:          false,
-				Error:            fmt.Sprintf("Image mismatch: checkpoint was from '%s' but placeholder uses '%s'. The placeholder must use the same base image. Use skip_image_validation=true to override.", checkpointMeta.Image, containerInfo.Image),
-				CheckpointImage:  checkpointMeta.Image,
-				PlaceholderImage: containerInfo.Image,
-			})
-			return
-		}
-		log.Printf("Image validation passed: checkpoint=%s, placeholder=%s", checkpointMeta.Image, containerInfo.Image)
-	}
-
-	// Write trigger file to placeholder's filesystem via /proc/<pid>/root
-	// The trigger file contains the checkpoint path
-	triggerPath := fmt.Sprintf("%s/%d/root/tmp/restore-trigger", s.config.HostProc, containerInfo.PID)
-
-	// Write the checkpoint path to the trigger file
-	if err := os.WriteFile(triggerPath, []byte(checkpointPath), 0644); err != nil {
-		writeJSON(w, http.StatusInternalServerError, TriggerRestoreResponse{
-			Success: false,
-			Error:   fmt.Sprintf("Failed to write trigger file: %v", err),
-		})
-		return
-	}
-
-	log.Printf("Triggered restore for placeholder %s (PID %d) from checkpoint %s",
-		req.PlaceholderContainerID, containerInfo.PID, req.CheckpointID)
-
-	writeJSON(w, http.StatusOK, TriggerRestoreResponse{
-		Success:          true,
-		Message:          fmt.Sprintf("Restore triggered for checkpoint %s", req.CheckpointID),
-		TriggerPath:      triggerPath,
-		CheckpointImage:  checkpointMeta.Image,
-		PlaceholderImage: containerInfo.Image,
-	})
-}
-
-func (s *Server) handleListCheckpoints(w http.ResponseWriter, r *http.Request) {
-	if r.Method != http.MethodGet {
-		http.Error(w, "Method not allowed", http.StatusMethodNotAllowed)
-		return
-	}
-
-	checkpointIDs, err := common.ListCheckpoints(s.config.CheckpointDir)
-	if err != nil {
-		writeJSON(w, http.StatusInternalServerError, map[string]string{
-			"error": err.Error(),
-		})
-		return
-	}
-
-	var checkpoints []CheckpointInfo
-	for _, id := range checkpointIDs {
-		meta, err := common.GetCheckpointInfo(s.config.CheckpointDir, id)
-		if err != nil {
-			continue
-		}
-		checkpoints = append(checkpoints, CheckpointInfo{
-			ID:           meta.CheckpointID,
-			CreatedAt:    meta.CreatedAt,
-			SourceNode:   meta.SourceNode,
-			ContainerID:  meta.ContainerID,
-			PodName:      meta.PodName,
-			PodNamespace: meta.PodNamespace,
-			Image:        meta.Image,
-		})
-	}
-
-	writeJSON(w, http.StatusOK, ListCheckpointsResponse{
-		Checkpoints: checkpoints,
-	})
-}
-
-func writeJSON(w http.ResponseWriter, status int, data interface{}) {
-	w.Header().Set("Content-Type", "application/json")
-	w.WriteHeader(status)
-	json.NewEncoder(w).Encode(data)
-}
-
-func loggingMiddleware(next http.Handler) http.Handler {
-	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		start := time.Now()
-		log.Printf("Started %s %s", r.Method, r.URL.Path)
-		next.ServeHTTP(w, r)
-		log.Printf("Completed %s %s in %v", r.Method, r.URL.Path, time.Since(start))
-	})
-}
-
-func getEnv(key, defaultValue string) string {
-	if value := os.Getenv(key); value != "" {
-		return value
-	}
-	return defaultValue
-}
-
-// imagesCompatible checks if two container images are compatible for CRIU restore.
-// The placeholder image must be based on the same image as the checkpoint.
-// Handles various image name formats:
-//   - nginx:alpine vs nginx:alpine (exact match)
-//   - docker.io/library/nginx:alpine vs nginx:alpine (registry prefix)
-//   - criu-placeholder-nginx-alpine vs nginx:alpine (placeholder naming convention)
-func imagesCompatible(checkpointImage, placeholderImage string) bool {
-	// Exact match
-	if checkpointImage == placeholderImage {
-		return true
-	}
-
-	// Normalize images by removing common registry prefixes
-	normalize := func(img string) string {
-		// Remove docker.io/library/ prefix
-		img = strings.TrimPrefix(img, "docker.io/library/")
-		// Remove docker.io/ prefix
-		img = strings.TrimPrefix(img, "docker.io/")
-		return img
-	}
-
-	normalizedCheckpoint := normalize(checkpointImage)
-	normalizedPlaceholder := normalize(placeholderImage)
-
-	if normalizedCheckpoint == normalizedPlaceholder {
-		return true
-	}
-
-	// Check if placeholder follows criu-placeholder-<image> naming convention
-	// e.g., criu-placeholder-nginx-alpine should match nginx:alpine
-	if strings.HasPrefix(normalizedPlaceholder, "criu-placeholder-") {
-		// Convert nginx:alpine to nginx-alpine for comparison
-		checkpointSanitized := strings.ReplaceAll(normalizedCheckpoint, ":", "-")
-		checkpointSanitized = strings.ReplaceAll(checkpointSanitized, "/", "-")
-		expectedPlaceholder := "criu-placeholder-" + checkpointSanitized
-
-		if normalizedPlaceholder == expectedPlaceholder ||
-			strings.HasPrefix(normalizedPlaceholder, expectedPlaceholder+":") {
-			return true
-		}
-	}
-
-	return false
+	log.Println("Agent stopped")
 }

deploy/chrek/cmd/restore-entrypoint/main.go

@@ -5,8 +5,12 @@ package main
 
 import (
 	"context"
+	"fmt"
 	"os"
+	"os/exec"
 	"os/signal"
+	"path/filepath"
+	"strings"
 	"syscall"
 
 	"github.com/sirupsen/logrus"
@@ -14,7 +18,50 @@ import (
 	"github.com/ai-dynamo/dynamo/deploy/chrek/pkg/restore"
 )
 
+// logGPUDiagnostics logs nvidia-smi output and /dev/nvidia* devices for debugging GPU visibility.
+func logGPUDiagnostics(label string) {
+	fmt.Printf("=== GPU DIAGNOSTICS [%s] ===\n", label)
+
+	// nvidia-smi
+	if out, err := exec.Command("nvidia-smi", "-L").CombinedOutput(); err != nil {
+		fmt.Printf("nvidia-smi -L: error: %v\n", err)
+	} else {
+		fmt.Printf("nvidia-smi -L:\n%s", out)
+	}
+
+	// GPU memory usage
+	if out, err := exec.Command("nvidia-smi", "--query-gpu=index,uuid,memory.used,memory.total,memory.free", "--format=csv,noheader").CombinedOutput(); err != nil {
+		fmt.Printf("nvidia-smi memory query: error: %v\n", err)
+	} else {
+		fmt.Printf("nvidia-smi memory:\n%s", out)
+	}
+
+	// /dev/nvidia* devices
+	matches, _ := filepath.Glob("/dev/nvidia*")
+	fmt.Printf("/dev/nvidia* devices: %s\n", strings.Join(matches, ", "))
+
+	// NVIDIA_VISIBLE_DEVICES env
+	fmt.Printf("NVIDIA_VISIBLE_DEVICES=%s\n", os.Getenv("NVIDIA_VISIBLE_DEVICES"))
+	fmt.Printf("CUDA_VISIBLE_DEVICES=%s\n", os.Getenv("CUDA_VISIBLE_DEVICES"))
+
+	// Linux namespaces for PID 1
+	for _, ns := range []string{"mnt", "pid", "ipc", "net", "uts", "cgroup"} {
+		link, err := os.Readlink(fmt.Sprintf("/proc/1/ns/%s", ns))
+		if err != nil {
+			link = err.Error()
+		}
+		fmt.Printf("ns/%s: %s\n", ns, link)
+	}
+
+	fmt.Printf("=== END GPU DIAGNOSTICS [%s] ===\n", label)
+}
+
 func main() {
+	// Log GPU diagnostics BEFORE anything else (gated on DEBUG for production quietness)
+	if os.Getenv("DEBUG") == "1" {
+		logGPUDiagnostics("PRE-RESTORE")
+	}
+
 	// Set up logging
 	log := logrus.New()
 	log.SetOutput(os.Stdout)
@@ -23,8 +70,12 @@ func main() {
 		TimestampFormat: "2006-01-02 15:04:05",
 	})
 
-	// Load configuration from environment
-	cfg := restore.ConfigFromEnv()
+	// Load configuration from hardcoded defaults + operator-injected env vars.
+	// os.Args[1:] are the cold start command args (passed by the operator via pod spec).
+	cfg, err := restore.NewRestoreRequest(os.Args[1:])
+	if err != nil {
+		log.WithError(err).Fatal("Failed to load restore configuration")
+	}
 
 	// Set log level based on DEBUG flag
 	if cfg.Debug {

deploy/chrek/pkg/checkpoint/checkpoint.go

@@ -6,19 +6,57 @@ import (
 	"fmt"
 	"os"
 	"path/filepath"
-	"strings"
 	"time"
 
-	criu "github.com/checkpoint-restore/go-criu/v7"
+	criurpc "github.com/checkpoint-restore/go-criu/v7/rpc"
+	specs "github.com/opencontainers/runtime-spec/specs-go"
 	"github.com/sirupsen/logrus"
 	"google.golang.org/protobuf/proto"
 
-	checkpointk8s "github.com/ai-dynamo/dynamo/deploy/chrek/pkg/checkpoint/k8s"
 	"github.com/ai-dynamo/dynamo/deploy/chrek/pkg/common"
 )
 
-// Options configures the checkpoint operation
-type Options struct {
+// ContainerInfoSnapshot holds runtime/container info needed for checkpointing.
+type ContainerInfoSnapshot struct {
+	PID        int
+	RootFS     string
+	UpperDir   string
+	OCISpec    *specs.Spec
+	MountInfo  []MountInfo
+	Namespaces map[NamespaceType]*NamespaceInfo
+}
+
+// CheckpointManifest is saved as manifest.yaml at checkpoint time and loaded at restore.
+type CheckpointManifest struct {
+	CheckpointID string    `yaml:"checkpointId"`
+	CreatedAt    time.Time `yaml:"createdAt"`
+
+	CRIUDump   CRIUDumpManifest         `yaml:"criuDump"`
+	K8s        SourcePodManifest        `yaml:"k8s"`
+	Filesystem FilesystemManifest       `yaml:"filesystem"`
+	Namespaces []NamespaceManifestEntry `yaml:"namespaces"`
+}
+
+// NewCheckpointManifest assembles a CheckpointManifest from per-module builders.
+func NewCheckpointManifest(
+	checkpointID string,
+	criuDump CRIUDumpManifest,
+	k8s SourcePodManifest,
+	filesystem FilesystemManifest,
+	namespaces []NamespaceManifestEntry,
+) *CheckpointManifest {
+	return &CheckpointManifest{
+		CheckpointID: checkpointID,
+		CreatedAt:    time.Now().UTC(),
+		CRIUDump:     criuDump,
+		K8s:          k8s,
+		Filesystem:   filesystem,
+		Namespaces:   namespaces,
+	}
+}
+
+// CheckpointRequest holds per-checkpoint identifiers for a checkpoint operation.
+type CheckpointRequest struct {
 	ContainerID   string
 	ContainerName string // K8s container name (for K8s API volume type lookup)
 	CheckpointID  string
@@ -26,237 +64,180 @@ type Options struct {
 	NodeName      string
 	PodName       string
 	PodNamespace  string
-
-	// CRIU options (from environment variables)
-	GhostLimit uint32 // From CRIU_GHOST_LIMIT: ghost file size limit in bytes (0 = CRIU default)
-	Timeout    uint32 // From CRIU_TIMEOUT: timeout in seconds (0 = no timeout)
-
-	// GPU/CUDA checkpoint options
-	CUDAPluginDir  string   // Path to CRIU CUDA plugin (e.g., /home/mmshin/work/criu/plugins/cuda)
-	ExternalMounts []string // Additional external mount mappings (e.g., "mnt[path]:path")
 }
 
-// Result contains the result of a checkpoint operation
-type Result struct {
+// CheckpointOutcome contains the result of a checkpoint operation.
+type CheckpointOutcome struct {
 	CheckpointID  string
 	CheckpointDir string
-	Metadata      *common.CheckpointMetadata
+	Data          *CheckpointManifest
 }
 
 // Checkpointer performs CRIU checkpoint operations
 type Checkpointer struct {
-	discoveryClient *checkpointk8s.DiscoveryClient
-	k8sClient       *checkpointk8s.K8sClient // Optional: for accurate volume type discovery from K8s API
-	hostProc        string
+	discoveryClient *DiscoveryClient
 	log             *logrus.Entry
 }
 
 // NewCheckpointer creates a new checkpointer
-func NewCheckpointer(discoveryClient *checkpointk8s.DiscoveryClient, hostProc string) *Checkpointer {
-	if hostProc == "" {
-		hostProc = os.Getenv("HOST_PROC")
-		if hostProc == "" {
-			hostProc = "/proc"
-		}
-	}
+func NewCheckpointer(discoveryClient *DiscoveryClient) *Checkpointer {
 	return &Checkpointer{
 		discoveryClient: discoveryClient,
-		hostProc:        hostProc,
 		log:             logrus.WithField("component", "checkpointer"),
 	}
 }
 
-// WithK8sClient sets an optional Kubernetes client for accurate volume type discovery.
-// When set, volume types are fetched from the K8s API instead of being inferred from paths.
-func (c *Checkpointer) WithK8sClient(client *checkpointk8s.K8sClient) *Checkpointer {
-	c.k8sClient = client
-	return c
-}
-
-// Checkpoint performs a CRIU dump of a container
-func (c *Checkpointer) Checkpoint(ctx context.Context, opts Options) (*Result, error) {
+// Checkpoint performs a CRIU dump of a container.
+// The operation has three phases: introspect, configure, capture.
+func (c *Checkpointer) Checkpoint(ctx context.Context, req CheckpointRequest, spec *CheckpointSpec) (*CheckpointOutcome, error) {
+	if spec == nil {
+		return nil, fmt.Errorf("checkpoint spec is required")
+	}
 	checkpointStart := time.Now()
 	c.log.Info("=== Starting checkpoint operation ===")
 
-	// 1. Resolve container to get PID
-	resolveStart := time.Now()
-	containerInfo, err := c.discoveryClient.ResolveContainer(ctx, opts.ContainerID)
-	if err != nil {
-		return nil, fmt.Errorf("failed to resolve container: %w", err)
-	}
-	pid := int(containerInfo.PID)
-	c.log.WithField("duration", time.Since(resolveStart)).Info("Container resolution completed")
-
-	// 2. Create checkpoint directory
-	checkpointDir := common.GetCheckpointDir(opts.CheckpointDir, opts.CheckpointID)
+	checkpointDir := filepath.Join(req.CheckpointDir, req.CheckpointID)
 	if err := os.MkdirAll(checkpointDir, 0700); err != nil {
 		return nil, fmt.Errorf("failed to create checkpoint directory: %w", err)
 	}
 
-	// 3. Introspect container state
-	introspectStart := time.Now()
-	rootFS, err := GetRootFS(pid, c.hostProc)
+	// Open image directory FD for CRIU — must stay open through both configure and capture
+	// phases since CRIU's swrk child process inherits this FD.
+	imageDir, imageDirFD, err := common.OpenPathForCRIU(checkpointDir)
 	if err != nil {
-		return nil, fmt.Errorf("failed to get rootfs: %w", err)
+		return nil, fmt.Errorf("failed to open image directory: %w", err)
 	}
-	mounts, err := GetKubernetesVolumeMounts(pid, c.hostProc)
+	defer imageDir.Close()
+
+	// Phase 1: Introspect container state
+	state, err := c.introspect(ctx, req.ContainerID)
 	if err != nil {
-		return nil, fmt.Errorf("failed to get mounts: %w", err)
+		return nil, err
 	}
-	namespaces, err := GetAllNamespaces(pid, c.hostProc)
+
+	// Phase 2: Configure CRIU options and build checkpoint manifest.
+	criuOpts, data, err := c.configure(state, req, spec, checkpointDir, imageDirFD)
 	if err != nil {
-		return nil, fmt.Errorf("failed to get namespaces: %w", err)
+		return nil, err
 	}
-	c.log.WithField("duration", time.Since(introspectStart)).Info("Container introspection completed")
 
-	// 4. Open image directory FD
-	imageDir, imageDirFD, err := OpenImageDir(checkpointDir)
+	// Phase 3: Capture — CRIU dump, /dev/shm, rootfs diff
+	criuDumpDuration, err := c.capture(criuOpts, data, state, checkpointDir)
 	if err != nil {
 		return nil, err
 	}
-	defer imageDir.Close()
 
-	// 5. Build CRIU options
-	criuOpts := BuildCRIUOptsFromCheckpointOpts(opts, pid, imageDirFD, rootFS)
+	totalDuration := time.Since(checkpointStart)
+	c.log.WithFields(logrus.Fields{
+		"total_duration":     totalDuration,
+		"criu_dump_duration": criuDumpDuration,
+	}).Info("=== Checkpoint operation completed ===")
 
-	// 6. Create CRIU config file for CUDA plugin (libdir is not available via RPC)
-	if opts.CUDAPluginDir != "" {
-		if opts.Timeout == 0 {
-			return nil, fmt.Errorf("CRIU_TIMEOUT environment variable must be set for CUDA checkpoints")
-		}
-		configPath := filepath.Join(checkpointDir, "criu.conf")
-		configContent := fmt.Sprintf(`enable-external-masters
-libdir %s
-tcp-close
-link-remap
-timeout %d
-allow-uprobes
-skip-in-flight
-`, opts.CUDAPluginDir, opts.Timeout)
-		if err := os.WriteFile(configPath, []byte(configContent), 0644); err != nil {
-			return nil, fmt.Errorf("failed to write CRIU config file: %w", err)
-		}
-		criuOpts.ConfigFile = proto.String(configPath)
-		c.log.WithFields(logrus.Fields{
-			"config_path": configPath,
-			"plugin_dir":  opts.CUDAPluginDir,
-		}).Info("Created CRIU config file for CUDA plugin")
-	}
+	return &CheckpointOutcome{
+		CheckpointID:  req.CheckpointID,
+		CheckpointDir: checkpointDir,
+		Data:          data,
+	}, nil
+}
 
-	// 7. Configure external mounts and namespaces
-	if err := ConfigureExternalMounts(criuOpts, pid, c.hostProc, containerInfo); err != nil {
-		return nil, err
-	}
-	netNsInode := ConfigureExternalNamespaces(criuOpts, namespaces, opts.ExternalMounts)
-	if netNsInode > 0 {
-		c.log.WithField("inode", netNsInode).Debug("Marked network namespace as external")
-	}
-	for _, extMount := range opts.ExternalMounts {
-		c.log.WithField("external", extMount).Debug("Added external mount mapping")
+// introspect resolves the container and gathers all runtime state from containerd and /proc.
+func (c *Checkpointer) introspect(ctx context.Context, containerID string) (*ContainerInfoSnapshot, error) {
+	pid, ociSpec, err := c.discoveryClient.ResolveContainer(ctx, containerID)
+	if err != nil {
+		return nil, fmt.Errorf("failed to resolve container: %w", err)
 	}
 
-	// 8. Get overlay upperdir for rootfs diff capture
-	upperDir, upperDirErr := GetOverlayUpperDir(pid, c.hostProc)
-	if upperDirErr != nil {
-		c.log.WithError(upperDirErr).Warn("Could not get overlay upperdir - rootfs diff will not be captured")
-	} else {
-		c.log.WithField("upperdir", upperDir).Debug("Found overlay upperdir")
+	rootFS, err := GetRootFS(pid)
+	if err != nil {
+		return nil, fmt.Errorf("failed to get rootfs: %w", err)
 	}
-
-	// 9. Build and save initial metadata before dump
-	metaCfg := MetadataBuilderConfig{
-		CheckpointID:  opts.CheckpointID,
-		NodeName:      opts.NodeName,
-		ContainerID:   opts.ContainerID,
-		ContainerName: opts.ContainerName,
-		PodName:       opts.PodName,
-		PodNamespace:  opts.PodNamespace,
-		PID:           pid,
-		CUDAPluginDir: opts.CUDAPluginDir,
+	upperDir, err := GetOverlayUpperDir(pid)
+	if err != nil {
+		return nil, fmt.Errorf("failed to get overlay upperdir: %w", err)
 	}
-	meta := BuildCheckpointMetadata(ctx, metaCfg, containerInfo, mounts, namespaces, c.k8sClient, c.log)
-	if upperDir != "" {
-		meta.UpperDir = upperDir
+	mountInfo, err := ReadMountInfoFromHostProcPath(pid)
+	if err != nil {
+		return nil, fmt.Errorf("failed to parse mountinfo: %w", err)
 	}
-	if err := common.SaveMetadata(checkpointDir, meta); err != nil {
-		return nil, fmt.Errorf("failed to save metadata: %w", err)
+	namespaces, err := GetAllNamespaces(pid)
+	if err != nil {
+		return nil, fmt.Errorf("failed to get namespaces: %w", err)
 	}
 
-	// 10. Remove semaphores from /dev/shm before checkpoint
-	// Semaphores cause CRIU restore to fail with "Can't link dev/shm/link_remap.X -> dev/shm/sem.Y"
-	if err := c.removeSemaphores(pid); err != nil {
-		return nil, fmt.Errorf("failed to remove semaphores: %w", err)
+	return &ContainerInfoSnapshot{
+		PID:        pid,
+		RootFS:     rootFS,
+		UpperDir:   upperDir,
+		OCISpec:    ociSpec,
+		MountInfo:  mountInfo,
+		Namespaces: namespaces,
+	}, nil
+}
+
+// configure builds CRIU options and checkpoint manifest from runtime snapshot and spec.
+func (c *Checkpointer) configure(
+	state *ContainerInfoSnapshot,
+	req CheckpointRequest,
+	spec *CheckpointSpec,
+	checkpointDir string,
+	imageDirFD int32,
+) (*criurpc.CriuOpts, *CheckpointManifest, error) {
+	criuOpts, err := BuildCRIUDumpOptions(
+		&spec.CRIU,
+		state.PID,
+		imageDirFD,
+		state.RootFS,
+		state.MountInfo,
+		state.OCISpec,
+		state.Namespaces,
+	)
+	if err != nil {
+		return nil, nil, err
 	}
 
-	// 11. Execute CRIU dump via go-criu
-	criuDumpStart := time.Now()
-	criuClient := criu.MakeCriu()
-	if err := criuClient.Dump(criuOpts, nil); err != nil {
-		c.log.WithField("duration", time.Since(criuDumpStart)).Error("CRIU dump failed")
-		return nil, fmt.Errorf("CRIU dump failed: %w", err)
+	// Write CRIU config file (for options unavailable via RPC)
+	configPath := filepath.Join(checkpointDir, CheckpointCRIUConfFilename)
+	if err := os.WriteFile(configPath, []byte(spec.CRIU.GenerateCRIUConfContent()), 0644); err != nil {
+		return nil, nil, fmt.Errorf("failed to write CRIU config file: %w", err)
 	}
-	criuDumpDuration := time.Since(criuDumpStart)
-	c.log.WithField("duration", criuDumpDuration).Info("CRIU dump completed successfully")
+	criuOpts.ConfigFile = proto.String(configPath)
 
-	// 12. Capture rootfs diff and deleted files
-	rootfsCaptureStart := time.Now()
-	CaptureRootfsState(upperDir, checkpointDir, meta, c.log)
-	c.log.WithField("duration", time.Since(rootfsCaptureStart)).Info("Rootfs capture completed")
+	// Build and save the checkpoint manifest.
+	manifest := NewCheckpointManifest(
+		req.CheckpointID,
+		NewCRIUDumpManifest(criuOpts, spec.CRIU),
+		NewSourcePodManifest(req, state.PID),
+		NewFilesystemManifest(spec.RootfsExclusions, state.UpperDir, state.OCISpec),
+		NewNamespaceManifestEntries(state.Namespaces),
+	)
 
-	totalDuration := time.Since(checkpointStart)
-	c.log.WithFields(logrus.Fields{
-		"total_duration":     totalDuration,
-		"criu_dump_duration": criuDumpDuration,
-	}).Info("=== Checkpoint operation completed ===")
+	if err := WriteCheckpointManifest(checkpointDir, manifest); err != nil {
+		return nil, nil, fmt.Errorf("failed to write checkpoint manifest: %w", err)
+	}
 
-	return &Result{
-		CheckpointID:  opts.CheckpointID,
-		CheckpointDir: checkpointDir,
-		Metadata:      meta,
-	}, nil
+	return criuOpts, manifest, nil
 }
 
-// removeSemaphores removes POSIX semaphores from the container's /dev/shm.
-// Semaphores can cause issues during CRIU checkpoint/restore because they
-// maintain kernel state that may not transfer correctly between processes.
-// This accesses the container's filesystem via /proc/<pid>/root/dev/shm/.
-func (c *Checkpointer) removeSemaphores(pid int) error {
-	shmPath := filepath.Join(c.hostProc, fmt.Sprintf("%d/root/dev/shm", pid))
-
-	entries, err := os.ReadDir(shmPath)
+// capture executes the CRIU dump and post-dump captures (/dev/shm, rootfs diff).
+// Returns the CRIU dump duration for timing reporting.
+func (c *Checkpointer) capture(
+	criuOpts *criurpc.CriuOpts,
+	data *CheckpointManifest,
+	state *ContainerInfoSnapshot,
+	checkpointDir string,
+) (time.Duration, error) {
+	criuDumpDuration, err := ExecuteCRIUDump(criuOpts, checkpointDir, c.log)
 	if err != nil {
-		// It's okay if /dev/shm doesn't exist (container may not have it)
-		c.log.WithError(err).Debug("Could not read container /dev/shm (may not exist)")
-		return nil
+		return 0, err
 	}
 
-	var removed []string
-	var errors []error
-	for _, entry := range entries {
-		name := entry.Name()
-		if strings.HasPrefix(name, "sem.") {
-			semPath := filepath.Join(shmPath, name)
-			if err := os.Remove(semPath); err != nil {
-				c.log.WithError(err).WithField("semaphore", name).Error("Failed to remove semaphore")
-				errors = append(errors, fmt.Errorf("failed to remove semaphore %s: %w", name, err))
-			} else {
-				removed = append(removed, name)
-			}
-		}
+	// Capture /dev/shm contents (must happen after dump for final process state)
+	if err := CaptureDevShm(state.PID, checkpointDir, c.log); err != nil {
+		c.log.WithError(err).Warn("Failed to capture /dev/shm contents")
 	}
 
-	if len(errors) > 0 {
-		return fmt.Errorf("failed to remove %d semaphore(s): %v", len(errors), errors)
-	}
-
-	if len(removed) > 0 {
-		c.log.WithFields(logrus.Fields{
-			"count":      len(removed),
-			"semaphores": removed,
-		}).Info("Removed semaphores from container /dev/shm before checkpoint")
-	} else {
-		c.log.Debug("No semaphores found in container /dev/shm")
-	}
+	// Capture rootfs diff and deleted files
+	CaptureRootfsState(state.UpperDir, checkpointDir, data, c.log)
 
-	return nil
+	return criuDumpDuration, nil
 }

deploy/chrek/pkg/checkpoint/config.go

+// config.go defines the static checkpoint spec loaded from ConfigMap YAML.
+package checkpoint
+
+import "fmt"
+
+// CheckpointSpec is the static checkpoint spec loaded from ConfigMap YAML.
+type CheckpointSpec struct {
+	// BasePath is the base directory for checkpoint storage (PVC mount point).
+	BasePath string `yaml:"basePath"`
+
+	// CRIU options for dump operations
+	CRIU CRIUSettings `yaml:"criu"`
+
+	// RootfsExclusions defines paths to exclude from rootfs diff capture
+	RootfsExclusions FilesystemConfig `yaml:"rootfsExclusions"`
+}
+
+// Validate checks that the CheckpointSpec has valid values.
+func (c *CheckpointSpec) Validate() error {
+	return c.RootfsExclusions.Validate()
+}
+
+// ConfigError represents a configuration validation error.
+type ConfigError struct {
+	Field   string
+	Message string
+}
+
+func (e *ConfigError) Error() string {
+	return fmt.Sprintf("config error: %s: %s", e.Field, e.Message)
+}

deploy/chrek/pkg/checkpoint/constants.go

+// constants.go defines shared constants used across checkpoint and restore packages.
+package checkpoint
+
+const (
+	// HostProcPath is the mount point for the host's /proc in DaemonSet pods.
+	HostProcPath = "/host/proc"
+
+	// DevShmDirName is the directory name for captured /dev/shm contents.
+	DevShmDirName = "dev-shm"
+
+	// KubeLabelCheckpointSource is the pod label that triggers automatic checkpointing.
+	// Set by the operator on checkpoint-eligible pods.
+	KubeLabelCheckpointSource = "nvidia.com/checkpoint-source"
+
+	// KubeLabelCheckpointHash is the pod label specifying the checkpoint identity hash.
+	// Set by the operator on checkpoint-eligible pods.
+	KubeLabelCheckpointHash = "nvidia.com/checkpoint-hash"
+
+	// DumpLogFilename is the CRIU dump (checkpoint) log filename.
+	DumpLogFilename = "dump.log"
+
+	// CheckpointCRIUConfFilename is the CRIU config file written at checkpoint time.
+	CheckpointCRIUConfFilename = "criu.conf"
+
+	// CheckpointDoneFilename is the marker file written to the checkpoint directory
+	// after all checkpoint artifacts are complete. Used to detect checkpoint readiness.
+	// Also hard-coded in vLLM for early-exit when checkpoint already exists.
+	CheckpointDoneFilename = "checkpoint.done"
+
+	// CheckpointManifestFilename is the name of the manifest file in checkpoint directories.
+	CheckpointManifestFilename = "manifest.yaml"
+
+	// DescriptorsFilename is the name of the file descriptors file.
+	DescriptorsFilename = "descriptors.yaml"
+
+	// RootfsDiffFilename is the name of the rootfs diff tar in checkpoint directories.
+	RootfsDiffFilename = "rootfs-diff.tar"
+
+	// DeletedFilesFilename is the name of the deleted files JSON in checkpoint directories.
+	DeletedFilesFilename = "deleted-files.json"
+)

deploy/chrek/pkg/checkpoint/criu.go

@@ -3,176 +3,256 @@ package checkpoint
 
 import (
 	"fmt"
-	"os"
+	"time"
 
+	criu "github.com/checkpoint-restore/go-criu/v7"
 	criurpc "github.com/checkpoint-restore/go-criu/v7/rpc"
+	specs "github.com/opencontainers/runtime-spec/specs-go"
+	"github.com/sirupsen/logrus"
 	"google.golang.org/protobuf/proto"
-
-	checkpointk8s "github.com/ai-dynamo/dynamo/deploy/chrek/pkg/checkpoint/k8s"
-	"github.com/ai-dynamo/dynamo/deploy/chrek/pkg/common"
 )
 
-// CRIUConfig holds configuration for CRIU dump operations.
-// Most options are always-on with safe defaults for K8s environments.
-type CRIUConfig struct {
-	PID        int
-	ImageDirFD int32
-	RootFS     string
-	GhostLimit uint32 // From env CRIU_GHOST_LIMIT: max ghost file size (0 = CRIU default)
-	Timeout    uint32 // From env CRIU_TIMEOUT: checkpoint timeout in seconds (0 = no timeout)
-}
+// CRIUSettings holds CRIU-specific configuration options.
+// Options are categorized by how they are passed to CRIU:
+//   - RPC options: Passed via go-criu CriuOpts protobuf
+//   - CRIU conf file options: Written to criu.conf (NOT available via RPC)
+type CRIUSettings struct {
+	// === RPC Options (passed via go-criu CriuOpts) ===
+
+	// GhostLimit is the maximum ghost file size in bytes.
+	// Ghost files are deleted-but-open files that CRIU needs to checkpoint.
+	// 512MB is recommended for GPU workloads with large memory allocations.
+	GhostLimit uint32 `yaml:"ghostLimit"`
+
+	// Timeout is the CRIU operation timeout in seconds.
+	// 6 hours (21600s) is recommended for large GPU model checkpoints.
+	Timeout uint32 `yaml:"timeout"`
+
+	// LogLevel is the CRIU logging verbosity (0-4).
+	LogLevel int32 `yaml:"logLevel"`
+
+	// WorkDir is the CRIU work directory for temporary files.
+	WorkDir string `yaml:"workDir"`
+
+	// AutoDedup enables auto-deduplication of memory pages.
+	AutoDedup bool `yaml:"autoDedup"`
+
+	// LazyPages enables lazy page migration (experimental).
+	LazyPages bool `yaml:"lazyPages"`
+
+	// LeaveRunning keeps the process running after checkpoint (dump only).
+	LeaveRunning bool `yaml:"leaveRunning"`
+
+	// ShellJob allows checkpointing session leaders (containers are often session leaders).
+	ShellJob bool `yaml:"shellJob"`
+
+	// TcpClose closes TCP connections instead of preserving them (pod IPs change on restore).
+	TcpClose bool `yaml:"tcpClose"`
+
+	// FileLocks allows checkpointing processes with file locks.
+	FileLocks bool `yaml:"fileLocks"`
+
+	// OrphanPtsMaster allows checkpointing containers with TTYs.
+	OrphanPtsMaster bool `yaml:"orphanPtsMaster"`
+
+	// ExtUnixSk allows external Unix sockets.
+	ExtUnixSk bool `yaml:"extUnixSk"`
 
-// OpenImageDir opens a checkpoint directory and prepares it for CRIU.
-// Returns the opened file and its FD. The caller must close the file when done.
-// The file descriptor has CLOEXEC cleared so it can be inherited by CRIU.
-func OpenImageDir(checkpointDir string) (*os.File, int32, error) {
-	return common.OpenDirForCRIU(checkpointDir)
+	// LinkRemap handles deleted-but-open files.
+	LinkRemap bool `yaml:"linkRemap"`
+
+	// ExtMasters allows external bind mount masters.
+	ExtMasters bool `yaml:"extMasters"`
+
+	// ManageCgroupsMode controls cgroup handling: "ignore" lets K8s manage cgroups.
+	ManageCgroupsMode string `yaml:"manageCgroupsMode"`
+
+	// === CRIU Conf File Options (NOT available via RPC - written to criu.conf) ===
+
+	// LibDir is the path to CRIU plugin directory (e.g., /usr/local/lib/criu).
+	// Required for CUDA checkpoint/restore.
+	LibDir string `yaml:"libDir"`
+
+	// AllowUprobes allows user-space probes (required for CUDA checkpoints).
+	AllowUprobes bool `yaml:"allowUprobes"`
+
+	// SkipInFlight skips in-flight TCP connections during checkpoint/restore.
+	SkipInFlight bool `yaml:"skipInFlight"`
 }
 
-// BuildCRIUOpts creates CRIU options from a config struct.
-// This sets up the base options; external mounts and namespaces are added separately.
-//
-// Always-on options for K8s:
-//   - LeaveRunning: always keep process running after checkpoint
-//   - ShellJob: containers are often session leaders
-//   - TcpClose: pod IPs change on restore/migration
-//   - FileLocks: applications use file locks
-//   - OrphanPtsMaster: containers with TTYs
-//   - ExtUnixSk: containers have external Unix sockets
-//   - ManageCgroups (IGNORE): let K8s manage cgroups
-//   - LinkRemap: handle deleted-but-open files (safe for all workloads)
-//   - ExtMasters: external bind mount masters (safe for all workloads)
-func BuildCRIUOpts(cfg CRIUConfig) *criurpc.CriuOpts {
-	cgMode := criurpc.CriuCgMode_IGNORE
-	criuOpts := &criurpc.CriuOpts{
-		Pid:               proto.Int32(int32(cfg.PID)),
-		ImagesDirFd:       proto.Int32(cfg.ImageDirFD),
-		LogLevel:          proto.Int32(4),
-		LogFile:           proto.String("dump.log"),
-		Root:              proto.String(cfg.RootFS),
-		ManageCgroups:     proto.Bool(true),
-		ManageCgroupsMode: &cgMode,
-		// Always-on for K8s environments
-		LeaveRunning:    proto.Bool(true),
-		ShellJob:        proto.Bool(true),
-		TcpClose:        proto.Bool(true),
-		FileLocks:       proto.Bool(true),
-		OrphanPtsMaster: proto.Bool(true),
-		ExtUnixSk:       proto.Bool(true),
-		LinkRemap:       proto.Bool(true),
-		ExtMasters:      proto.Bool(true),
-	}
+// GenerateCRIUConfContent generates the criu.conf file content for options
+// that cannot be passed via RPC.
+func (c *CRIUSettings) GenerateCRIUConfContent() string {
+	var content string
 
-	// Optional: ghost limit from env (0 = use CRIU default)
-	if cfg.GhostLimit > 0 {
-		criuOpts.GhostLimit = proto.Uint32(cfg.GhostLimit)
+	if c.LibDir != "" {
+		content += "libdir " + c.LibDir + "\n"
 	}
-
-	// Optional: timeout from env (0 = no timeout)
-	if cfg.Timeout > 0 {
-		criuOpts.Timeout = proto.Uint32(cfg.Timeout)
+	if c.AllowUprobes {
+		content += "allow-uprobes\n"
+	}
+	if c.SkipInFlight {
+		content += "skip-in-flight\n"
 	}
 
-	return criuOpts
+	return content
 }
 
-// AddExternalMounts adds mount points as external mounts to CRIU options.
-// CRIU requires all mounts to be marked as external for successful restore.
-func AddExternalMounts(criuOpts *criurpc.CriuOpts, mounts []AllMountInfo) {
-	addedMounts := make(map[string]bool)
+// ExternalMountManifestEntry is a serializable CRIU ext-mount entry in checkpoint manifests.
+type ExternalMountManifestEntry struct {
+	Key string `yaml:"key"`
+	Val string `yaml:"val"`
+}
 
-	for _, m := range mounts {
-		if addedMounts[m.MountPoint] {
-			continue
-		}
-		criuOpts.ExtMnt = append(criuOpts.ExtMnt, &criurpc.ExtMountMap{
-			Key: proto.String(m.MountPoint),
-			Val: proto.String(m.MountPoint),
-		})
-		addedMounts[m.MountPoint] = true
-	}
+// CRIUDumpManifest stores the resolved dump-time CRIU mount plan used for restore.
+type CRIUDumpManifest struct {
+	CRIU     CRIUSettings                 `yaml:"criu"`
+	ExtMnt   []ExternalMountManifestEntry `yaml:"extMnt,omitempty"`
+	External []string                     `yaml:"external,omitempty"`
+	SkipMnt  []string                     `yaml:"skipMnt,omitempty"`
 }
 
-// AddExternalPaths adds additional paths (masked/readonly) as external mounts.
-// These may not appear in mountinfo but CRIU still needs them marked as external.
-func AddExternalPaths(criuOpts *criurpc.CriuOpts, paths []string) {
-	// Build set of existing mount points
-	existing := make(map[string]bool)
-	for _, m := range criuOpts.ExtMnt {
-		existing[m.GetKey()] = true
+// NewCRIUDumpManifest serializes resolved dump options for restore.
+func NewCRIUDumpManifest(criuOpts *criurpc.CriuOpts, settings CRIUSettings) CRIUDumpManifest {
+	manifest := CRIUDumpManifest{CRIU: settings}
+	if criuOpts == nil {
+		return manifest
 	}
 
-	for _, path := range paths {
-		if existing[path] {
+	for _, mount := range criuOpts.ExtMnt {
+		if mount == nil || mount.GetKey() == "" {
 			continue
 		}
-		criuOpts.ExtMnt = append(criuOpts.ExtMnt, &criurpc.ExtMountMap{
-			Key: proto.String(path),
-			Val: proto.String(path),
+		manifest.ExtMnt = append(manifest.ExtMnt, ExternalMountManifestEntry{
+			Key: mount.GetKey(),
+			Val: mount.GetVal(),
 		})
-		existing[path] = true
 	}
+	manifest.External = append([]string(nil), criuOpts.External...)
+	manifest.SkipMnt = append([]string(nil), criuOpts.SkipMnt...)
+	return manifest
 }
 
-// AddExternalNamespace adds a namespace as external to CRIU options.
-// Format: "<type>[<inode>]:<key>"
-func AddExternalNamespace(criuOpts *criurpc.CriuOpts, nsType NamespaceType, inode uint64, key string) {
-	extNs := fmt.Sprintf("%s[%d]:%s", nsType, inode, key)
-	criuOpts.External = append(criuOpts.External, extNs)
-}
+// BuildCRIUDumpOptions creates CRIU options directly from spec settings and runtime state.
+func BuildCRIUDumpOptions(
+	settings *CRIUSettings,
+	pid int,
+	imageDirFD int32,
+	rootFS string,
+	mountInfo []MountInfo,
+	ociSpec *specs.Spec,
+	namespaces map[NamespaceType]*NamespaceInfo,
+) (*criurpc.CriuOpts, error) {
+	mountPolicy := BuildMountPolicy(mountInfo, ociSpec, rootFS)
 
-// AddExternalStrings adds raw external strings to CRIU options.
-// Used for additional external mount mappings (e.g., NVIDIA firmware files).
-func AddExternalStrings(criuOpts *criurpc.CriuOpts, externals []string) {
-	criuOpts.External = append(criuOpts.External, externals...)
-}
+	extMnt := buildExternalMountMaps(mountPolicy.Externalized)
+	skipMnt := mountPolicy.Skipped
+	external := buildExternalNamespaces(namespaces)
+	logrus.WithFields(logrus.Fields{
+		"externalized_count": len(mountPolicy.Externalized),
+		"skipped_count":      len(mountPolicy.Skipped),
+	}).Debug("Resolved mount policy for CRIU dump")
+
+	criuOpts := &criurpc.CriuOpts{
+		Pid:         proto.Int32(int32(pid)),
+		ImagesDirFd: proto.Int32(imageDirFD),
+		Root:        proto.String(rootFS),
+		LogFile:     proto.String(DumpLogFilename),
+	}
+	criuOpts.ExtMnt = extMnt
+	criuOpts.External = external
+	criuOpts.SkipMnt = skipMnt
+
+	if settings == nil {
+		return criuOpts, nil
+	}
+
+	// RPC options from spec.
+	criuOpts.LogLevel = proto.Int32(settings.LogLevel)
+	criuOpts.LeaveRunning = proto.Bool(settings.LeaveRunning)
+	criuOpts.ShellJob = proto.Bool(settings.ShellJob)
+	criuOpts.TcpClose = proto.Bool(settings.TcpClose)
+	criuOpts.FileLocks = proto.Bool(settings.FileLocks)
+	criuOpts.OrphanPtsMaster = proto.Bool(settings.OrphanPtsMaster)
+	criuOpts.ExtUnixSk = proto.Bool(settings.ExtUnixSk)
+	criuOpts.LinkRemap = proto.Bool(settings.LinkRemap)
+	criuOpts.ExtMasters = proto.Bool(settings.ExtMasters)
+	criuOpts.AutoDedup = proto.Bool(settings.AutoDedup)
+	criuOpts.LazyPages = proto.Bool(settings.LazyPages)
+
+	// Cgroup management mode
+	criuOpts.ManageCgroups = proto.Bool(true)
+	cgMode := criurpc.CriuCgMode_IGNORE
+	switch settings.ManageCgroupsMode {
+	case "soft":
+		cgMode = criurpc.CriuCgMode_SOFT
+	case "full":
+		cgMode = criurpc.CriuCgMode_FULL
+	case "strict":
+		cgMode = criurpc.CriuCgMode_STRICT
+	}
+	criuOpts.ManageCgroupsMode = &cgMode
 
-// ConfigureExternalMounts adds all required external mounts to CRIU options.
-// This includes mounts from /proc/pid/mountinfo plus masked/readonly paths from OCI spec.
-func ConfigureExternalMounts(criuOpts *criurpc.CriuOpts, pid int, hostProc string, containerInfo *checkpointk8s.ContainerInfo) error {
-	// Get all mounts from mountinfo - CRIU needs every mount marked as external
-	allMounts, err := GetAllMountsFromMountinfo(pid, hostProc)
-	if err != nil {
-		return fmt.Errorf("failed to get all mounts from mountinfo: %w", err)
+	// Optional numeric options
+	if settings.GhostLimit > 0 {
+		criuOpts.GhostLimit = proto.Uint32(settings.GhostLimit)
+	}
+	if settings.Timeout > 0 {
+		criuOpts.Timeout = proto.Uint32(settings.Timeout)
 	}
 
-	// Add mounts from mountinfo
-	AddExternalMounts(criuOpts, allMounts)
+	return criuOpts, nil
+}
 
-	// Add masked and readonly paths from OCI spec
-	AddExternalPaths(criuOpts, containerInfo.GetMaskedPaths())
-	AddExternalPaths(criuOpts, containerInfo.GetReadonlyPaths())
+// buildExternalMountMaps serializes externalized mount paths into CRIU map entries.
+func buildExternalMountMaps(paths []string) []*criurpc.ExtMountMap {
+	extMnt := make([]*criurpc.ExtMountMap, 0, len(paths))
+	existing := make(map[string]struct{}, len(paths))
+	for _, path := range paths {
+		if path == "" {
+			continue
+		}
+		if _, ok := existing[path]; ok {
+			continue
+		}
+		extMnt = append(extMnt, &criurpc.ExtMountMap{
+			Key: proto.String(path),
+			Val: proto.String(path),
+		})
+		existing[path] = struct{}{}
+	}
 
-	return nil
+	return extMnt
 }
 
-// ConfigureExternalNamespaces adds external namespaces to CRIU options.
-// Returns the network namespace inode if found, for logging purposes.
-func ConfigureExternalNamespaces(criuOpts *criurpc.CriuOpts, namespaces map[NamespaceType]*NamespaceInfo, externalMounts []string) uint64 {
-	var netNsInode uint64
+// buildExternalNamespaces builds external namespace/mount references.
+func buildExternalNamespaces(namespaces map[NamespaceType]*NamespaceInfo) []string {
+	external := make([]string, 0, 1)
 
 	// Mark network namespace as external for socket binding preservation
 	if netNs, ok := namespaces[NamespaceNet]; ok {
-		AddExternalNamespace(criuOpts, NamespaceNet, netNs.Inode, "extNetNs")
-		netNsInode = netNs.Inode
+		external = append(external, fmt.Sprintf("%s[%d]:%s", NamespaceNet, netNs.Inode, "extNetNs"))
+		logrus.WithField("inode", netNs.Inode).Debug("Marked network namespace as external")
 	}
 
-	// Add additional external mounts (e.g., for NVIDIA firmware files)
-	AddExternalStrings(criuOpts, externalMounts)
-
-	return netNsInode
+	return external
 }
 
-// BuildCRIUOptsFromCheckpointOpts constructs CRIU options from checkpoint Options.
-// Returns the configured CriuOpts ready for external mount/namespace configuration.
-func BuildCRIUOptsFromCheckpointOpts(opts Options, pid int, imageDirFD int32, rootFS string) *criurpc.CriuOpts {
-	cfg := CRIUConfig{
-		PID:        pid,
-		ImageDirFD: imageDirFD,
-		RootFS:     rootFS,
-		GhostLimit: opts.GhostLimit,
-		Timeout:    opts.Timeout,
+// ExecuteCRIUDump runs the CRIU dump and logs timing plus dump-log location on failure.
+func ExecuteCRIUDump(criuOpts *criurpc.CriuOpts, checkpointDir string, log *logrus.Entry) (time.Duration, error) {
+	criuDumpStart := time.Now()
+	criuClient := criu.MakeCriu()
+	if err := criuClient.Dump(criuOpts, nil); err != nil {
+		dumpDuration := time.Since(criuDumpStart)
+		log.WithFields(logrus.Fields{
+			"duration":       dumpDuration,
+			"checkpoint_dir": checkpointDir,
+			"dump_log_path":  fmt.Sprintf("%s/%s", checkpointDir, DumpLogFilename),
+		}).Error("CRIU dump failed")
+		return 0, fmt.Errorf("CRIU dump failed: %w", err)
 	}
 
-	return BuildCRIUOpts(cfg)
+	criuDumpDuration := time.Since(criuDumpStart)
+	log.WithField("duration", criuDumpDuration).Info("CRIU dump completed")
+	return criuDumpDuration, nil
 }

deploy/chrek/pkg/checkpoint/rootfs.go → deploy/chrek/pkg/checkpoint/filesystem.go

-// rootfs provides container rootfs introspection via /proc for CRIU checkpoint.
+// filesystem.go provides container rootfs introspection, filesystem config/metadata types,
+// and rootfs diff capture for CRIU checkpoint.
 package checkpoint
 
 import (
-	"bufio"
 	"encoding/json"
 	"fmt"
 	"os"
@@ -10,131 +10,152 @@ import (
 	"path/filepath"
 	"strings"
 
+	specs "github.com/opencontainers/runtime-spec/specs-go"
 	"github.com/sirupsen/logrus"
-
-	"github.com/ai-dynamo/dynamo/deploy/chrek/pkg/common"
 )
 
-// GetRootFS returns the container's root filesystem path
-// For containers using overlayfs, this extracts the upperdir
-func GetRootFS(pid int, hostProc string) (string, error) {
-	if hostProc == "" {
-		hostProc = "/proc"
-	}
+// FilesystemConfig is the static config for rootfs exclusions (from values.yaml).
+type FilesystemConfig struct {
+	// SystemDirs are system directories that should be excluded from rootfs diff.
+	// These directories are typically injected/bind-mounted by NVIDIA GPU Operator
+	// at container start time, so they already exist in the restore target.
+	// Excluding them prevents conflicts (especially socket files which cannot be overwritten).
+	// Default: ["./usr", "./etc", "./opt", "./var", "./run"]
+	SystemDirs []string `yaml:"systemDirs"`
+
+	// CacheDirs are cache directories that can safely be excluded to reduce checkpoint size.
+	// Model weights and other cached data are typically re-downloaded if needed.
+	// Default: ["./.cache/huggingface", "./.cache/torch"]
+	CacheDirs []string `yaml:"cacheDirs"`
+
+	// AdditionalExclusions are custom paths to exclude from the rootfs diff.
+	// Use this for application-specific exclusions.
+	// Paths should be relative with "./" prefix (e.g., "./data/temp").
+	AdditionalExclusions []string `yaml:"additionalExclusions"`
+}
 
-	// The rootfs is accessible via /proc/<pid>/root
-	// But for CRIU, we need the actual filesystem path
-	rootPath := fmt.Sprintf("%s/%d/root", hostProc, pid)
+// GetAllExclusions returns all exclusion paths combined.
+// This is used when building tar arguments for rootfs diff capture.
+func (c *FilesystemConfig) GetAllExclusions() []string {
+	if c == nil {
+		return nil
+	}
+	total := len(c.SystemDirs) + len(c.CacheDirs) + len(c.AdditionalExclusions)
+	exclusions := make([]string, 0, total)
+	exclusions = append(exclusions, c.SystemDirs...)
+	exclusions = append(exclusions, c.CacheDirs...)
+	exclusions = append(exclusions, c.AdditionalExclusions...)
+	return exclusions
+}
 
-	// Verify it exists
-	if _, err := os.Stat(rootPath); err != nil {
-		return "", fmt.Errorf("rootfs not accessible at %s: %w", rootPath, err)
+// Validate checks that the FilesystemConfig has valid values.
+func (c *FilesystemConfig) Validate() error {
+	if c == nil {
+		return nil
 	}
+	// All paths should start with "./" for tar relative path handling
+	for _, path := range c.GetAllExclusions() {
+		if !strings.HasPrefix(path, "./") {
+			return &ConfigError{
+				Field:   "rootfsExclusions",
+				Message: "all exclusion paths must start with './' (got: " + path + ")",
+			}
+		}
+	}
+	return nil
+}
 
-	return rootPath, nil
+// FilesystemManifest holds runtime filesystem state captured at checkpoint time.
+type FilesystemManifest struct {
+	Exclusions      FilesystemConfig `yaml:"exclusions"`
+	UpperDir        string           `yaml:"upperDir,omitempty"`
+	ExternalPaths   []string         `yaml:"externalPaths,omitempty"`
+	BindMountDests  []string         `yaml:"bindMountDests,omitempty"`
+	HasRootfsDiff   bool             `yaml:"hasRootfsDiff"`
+	HasDeletedFiles bool             `yaml:"hasDeletedFiles"`
 }
 
-// GetOverlayUpperDir extracts the overlay upperdir from mountinfo
-// This is the writable layer of the container's filesystem
-func GetOverlayUpperDir(pid int, hostProc string) (string, error) {
-	if hostProc == "" {
-		hostProc = "/proc"
+// NewFilesystemManifest constructs FilesystemManifest from config, overlay state, and OCI spec.
+func NewFilesystemManifest(exclusions FilesystemConfig, upperDir string, ociSpec *specs.Spec) FilesystemManifest {
+	meta := FilesystemManifest{
+		Exclusions: exclusions,
+		UpperDir:   upperDir,
 	}
-
-	mountinfoPath := fmt.Sprintf("%s/%d/mountinfo", hostProc, pid)
-	file, err := os.Open(mountinfoPath)
-	if err != nil {
-		return "", fmt.Errorf("failed to open mountinfo: %w", err)
+	if ociSpec == nil {
+		return meta
 	}
-	defer file.Close()
 
-	scanner := bufio.NewScanner(file)
-	for scanner.Scan() {
-		line := scanner.Text()
-		fields := strings.Fields(line)
-
-		// Look for the root mount (mount point is /)
-		// mountinfo format: id parent major:minor root mount-point options ... - fstype source super-options
-		if len(fields) < 5 {
-			continue
+	if ociSpec.Linux != nil {
+		meta.ExternalPaths = make([]string, 0, len(ociSpec.Linux.MaskedPaths)+len(ociSpec.Linux.ReadonlyPaths))
+		meta.ExternalPaths = append(meta.ExternalPaths, ociSpec.Linux.MaskedPaths...)
+		meta.ExternalPaths = append(meta.ExternalPaths, ociSpec.Linux.ReadonlyPaths...)
+	}
+	for _, m := range ociSpec.Mounts {
+		if m.Type == "bind" {
+			meta.BindMountDests = append(meta.BindMountDests, m.Destination)
 		}
+	}
+	return meta
+}
 
-		mountPoint := fields[4]
-		if mountPoint != "/" {
-			continue
-		}
+// GetRootFS returns the container's root filesystem path.
+func GetRootFS(pid int) (string, error) {
+	rootPath := fmt.Sprintf("%s/%d/root", HostProcPath, pid)
 
-		// Find the separator (-) to get fstype and options
-		sepIdx := -1
-		for i, f := range fields {
-			if f == "-" {
-				sepIdx = i
-				break
-			}
-		}
+	if _, err := os.Stat(rootPath); err != nil {
+		return "", fmt.Errorf("rootfs not accessible at %s: %w", rootPath, err)
+	}
 
-		if sepIdx == -1 || sepIdx+2 >= len(fields) {
-			continue
-		}
+	return rootPath, nil
+}
 
-		fsType := fields[sepIdx+1]
-		if fsType != "overlay" {
+// GetOverlayUpperDir extracts the overlay upperdir from mountinfo.
+// This is the writable layer of the container's filesystem.
+func GetOverlayUpperDir(pid int) (string, error) {
+	mountInfo, err := ReadMountInfoFromHostProcPath(pid)
+	if err != nil {
+		return "", fmt.Errorf("failed to parse mountinfo: %w", err)
+	}
+
+	for _, mount := range mountInfo {
+		if mount.MountPoint != "/" || mount.FSType != "overlay" {
 			continue
 		}
 
 		// Parse super options to find upperdir
-		superOptions := fields[sepIdx+3]
-		for _, opt := range strings.Split(superOptions, ",") {
+		for _, opt := range strings.Split(mount.SuperOptions, ",") {
 			if strings.HasPrefix(opt, "upperdir=") {
 				return strings.TrimPrefix(opt, "upperdir="), nil
 			}
 		}
 	}
 
-	if err := scanner.Err(); err != nil {
-		return "", fmt.Errorf("error reading mountinfo: %w", err)
-	}
-
 	return "", fmt.Errorf("overlay upperdir not found for pid %d", pid)
 }
 
-// DefaultRootfsDiffExclusions are paths excluded from the rootfs diff capture.
-// These directories are injected/bind-mounted by NVIDIA GPU Operator at container
-// start time, so they already exist in the restore target and cause conflicts
-// (especially socket files which cannot be overwritten).
-var DefaultRootfsDiffExclusions = []string{
-	// NVIDIA GPU Operator injects drivers, libraries, and config here
-	"./usr",
-	"./etc",
-	"./opt",
-	"./var",
-
-	// NVIDIA GPU Operator creates runtime sockets and firmware mounts here
-	// Socket files cause fatal tar errors even with --keep-old-files
-	"./run",
-}
-
 // CaptureRootfsDiff captures the overlay upperdir to a tar file.
 // The upperdir contains all filesystem modifications made by the container.
-// Excludes bind mount destinations and system directories to avoid conflicts during restore.
+// Excludes bind mount destinations and configured directories to avoid conflicts during restore.
 // Returns the path to the tar file or empty string if capture failed.
-func CaptureRootfsDiff(upperDir, checkpointDir string, excludePaths []string) (string, error) {
+func CaptureRootfsDiff(upperDir, checkpointDir string, exclusions *FilesystemConfig, bindMountDests []string) (string, error) {
 	if upperDir == "" {
 		return "", fmt.Errorf("upperdir is empty")
 	}
 
-	rootfsDiffPath := filepath.Join(checkpointDir, "rootfs-diff.tar")
+	rootfsDiffPath := filepath.Join(checkpointDir, RootfsDiffFilename)
 
 	// Build tar arguments with xattrs and exclusions
 	tarArgs := []string{"--xattrs"}
 
-	// Add default exclusions for system directories and caches
-	for _, excl := range DefaultRootfsDiffExclusions {
-		tarArgs = append(tarArgs, "--exclude="+excl)
+	// Add configured exclusions (systemDirs, cacheDirs, additionalExclusions from values.yaml)
+	if exclusions != nil {
+		for _, excl := range exclusions.GetAllExclusions() {
+			tarArgs = append(tarArgs, "--exclude="+excl)
+		}
 	}
 
 	// Add bind mount exclusions passed from caller
-	for _, dest := range excludePaths {
+	for _, dest := range bindMountDests {
 		// Convert absolute path to relative for tar (e.g., /etc/hosts -> ./etc/hosts)
 		tarArgs = append(tarArgs, "--exclude=."+dest)
 	}
@@ -165,7 +186,7 @@ func CaptureDeletedFiles(upperDir, checkpointDir string) (bool, error) {
 		return false, nil
 	}
 
-	deletedFilesPath := filepath.Join(checkpointDir, "deleted-files.json")
+	deletedFilesPath := filepath.Join(checkpointDir, DeletedFilesFilename)
 	data, err := json.Marshal(whiteouts)
 	if err != nil {
 		return false, fmt.Errorf("failed to marshal whiteouts: %w", err)
@@ -195,11 +216,11 @@ func FindWhiteoutFiles(upperDir string) ([]string, error) {
 			relPath, _ := filepath.Rel(upperDir, path)
 			dir := filepath.Dir(relPath)
 			deletedFile := strings.TrimPrefix(name, ".wh.")
-			if dir == "." {
-				whiteouts = append(whiteouts, deletedFile)
-			} else {
-				whiteouts = append(whiteouts, filepath.Join(dir, deletedFile))
+			deletedPath := deletedFile
+			if dir != "." {
+				deletedPath = filepath.Join(dir, deletedFile)
 			}
+			whiteouts = append(whiteouts, deletedPath)
 		}
 		return nil
 	})
@@ -208,23 +229,23 @@ func FindWhiteoutFiles(upperDir string) ([]string, error) {
 }
 
 // CaptureRootfsState captures the overlay upperdir and deleted files after CRIU dump.
-// Updates the metadata with rootfs diff information and saves it.
-func CaptureRootfsState(upperDir, checkpointDir string, meta *common.CheckpointMetadata, log *logrus.Entry) {
-	if upperDir == "" {
+// Updates the checkpoint manifest with rootfs diff information and saves it.
+func CaptureRootfsState(upperDir, checkpointDir string, data *CheckpointManifest, log *logrus.Entry) {
+	if upperDir == "" || data == nil {
 		return
 	}
 
-	// Capture rootfs diff
+	// Capture rootfs diff using exclusions from the checkpoint manifest.
+	configuredExclusions := data.Filesystem.Exclusions.GetAllExclusions()
 	log.WithFields(logrus.Fields{
-		"default_exclusions":    DefaultRootfsDiffExclusions,
-		"bind_mount_exclusions": meta.BindMountDests,
+		"configured_exclusions": configuredExclusions,
+		"bind_mount_exclusions": data.Filesystem.BindMountDests,
 	}).Debug("Rootfs diff exclusions")
-	rootfsDiffPath, err := CaptureRootfsDiff(upperDir, checkpointDir, meta.BindMountDests)
+	rootfsDiffPath, err := CaptureRootfsDiff(upperDir, checkpointDir, &data.Filesystem.Exclusions, data.Filesystem.BindMountDests)
 	if err != nil {
 		log.WithError(err).Warn("Failed to capture rootfs diff")
 	} else {
-		meta.RootfsDiffPath = rootfsDiffPath
-		meta.HasRootfsDiff = true
+		data.Filesystem.HasRootfsDiff = true
 		log.WithFields(logrus.Fields{
 			"upperdir": upperDir,
 			"tar_path": rootfsDiffPath,
@@ -236,12 +257,12 @@ func CaptureRootfsState(upperDir, checkpointDir string, meta *common.CheckpointM
 	if err != nil {
 		log.WithError(err).Warn("Failed to capture deleted files")
 	} else if hasDeletedFiles {
-		meta.HasDeletedFiles = true
+		data.Filesystem.HasDeletedFiles = true
 		log.Info("Recorded deleted files (whiteouts)")
 	}
 
-	// Update metadata with rootfs diff info
-	if err := common.SaveMetadata(checkpointDir, meta); err != nil {
-		log.WithError(err).Warn("Failed to update metadata with rootfs diff info")
+	// Update checkpoint manifest with rootfs diff info.
+	if err := WriteCheckpointManifest(checkpointDir, data); err != nil {
+		log.WithError(err).Warn("Failed to update checkpoint manifest with rootfs diff info")
 	}
 }

deploy/chrek/pkg/checkpoint/k8s.go

+// k8s contains containerd discovery and Kubernetes path classification helpers.
+package checkpoint
+
+import (
+	"context"
+	"fmt"
+
+	"github.com/containerd/containerd"
+	"github.com/containerd/containerd/namespaces"
+	specs "github.com/opencontainers/runtime-spec/specs-go"
+)
+
+const (
+	// K8sNamespace is the containerd namespace used by Kubernetes.
+	K8sNamespace = "k8s.io"
+
+	// ContainerdSocket is the default containerd socket path.
+	ContainerdSocket = "/run/containerd/containerd.sock"
+)
+
+type SourcePodManifest struct {
+	ContainerID  string `yaml:"containerId"`
+	PID          int    `yaml:"pid"`
+	SourceNode   string `yaml:"sourceNode"`
+	PodName      string `yaml:"podName"`
+	PodNamespace string `yaml:"podNamespace"`
+}
+
+func NewSourcePodManifest(params CheckpointRequest, pid int) SourcePodManifest {
+	return SourcePodManifest{
+		ContainerID:  params.ContainerID,
+		PID:          pid,
+		SourceNode:   params.NodeName,
+		PodName:      params.PodName,
+		PodNamespace: params.PodNamespace,
+	}
+}
+
+type DiscoveryClient struct {
+	client *containerd.Client
+}
+
+func NewDiscoveryClient() (*DiscoveryClient, error) {
+	client, err := containerd.New(ContainerdSocket)
+	if err != nil {
+		return nil, fmt.Errorf("failed to connect to containerd at %s: %w", ContainerdSocket, err)
+	}
+	return &DiscoveryClient{client: client}, nil
+}
+
+func (c *DiscoveryClient) Close() error {
+	if c.client != nil {
+		return c.client.Close()
+	}
+	return nil
+}
+
+func (c *DiscoveryClient) ResolveContainer(ctx context.Context, containerID string) (int, *specs.Spec, error) {
+	ctx = namespaces.WithNamespace(ctx, K8sNamespace)
+
+	container, err := c.client.LoadContainer(ctx, containerID)
+	if err != nil {
+		return 0, nil, fmt.Errorf("failed to load container %s: %w", containerID, err)
+	}
+
+	task, err := container.Task(ctx, nil)
+	if err != nil {
+		return 0, nil, fmt.Errorf("failed to get task for container %s: %w", containerID, err)
+	}
+
+	pid := task.Pid()
+
+	spec, err := container.Spec(ctx)
+	if err != nil {
+		return 0, nil, fmt.Errorf("failed to get spec for container %s: %w", containerID, err)
+	}
+
+	return int(pid), spec, nil
+}

deploy/chrek/pkg/checkpoint/k8s/discovery.go

-// discovery provides container information resolution via containerd.
-// This prefers containerd RPCs for configuration over /proc inspection,
-// following the principle that configuration should come from the container runtime
-// while runtime state (like namespace inodes) requires /proc.
-package k8s
-
-import (
-	"context"
-	"fmt"
-	"os"
-	"path/filepath"
-
-	"github.com/containerd/containerd"
-	"github.com/containerd/containerd/namespaces"
-	specs "github.com/opencontainers/runtime-spec/specs-go"
-)
-
-const (
-	// K8sNamespace is the containerd namespace used by Kubernetes
-	K8sNamespace = "k8s.io"
-	// DefaultSocket is the default containerd socket path
-	DefaultSocket = "/run/containerd/containerd.sock"
-)
-
-// ContainerInfo holds resolved container information from containerd.
-// Configuration data comes from containerd RPCs, runtime state from /proc.
-type ContainerInfo struct {
-	ContainerID string
-	PID         uint32
-	RootFS      string // Actual rootfs path (bundle path + spec.Root.Path)
-	BundlePath  string // Path to container bundle directory
-	Image       string
-	Spec        *specs.Spec // OCI spec from containerd (mounts, namespaces config)
-	Labels      map[string]string
-}
-
-// MountInfo represents a mount from the OCI spec.
-type MountInfo struct {
-	Destination string   // Mount point inside container
-	Source      string   // Source path on host
-	Type        string   // Filesystem type (bind, tmpfs, etc.)
-	Options     []string // Mount options
-}
-
-// NamespaceConfig represents namespace configuration from OCI spec.
-type NamespaceConfig struct {
-	Type string // Namespace type (network, pid, mount, etc.)
-	Path string // Path to namespace (empty for new namespace)
-}
-
-// DiscoveryClient wraps the containerd client for container discovery.
-type DiscoveryClient struct {
-	client *containerd.Client
-	socket string
-}
-
-// NewDiscoveryClient creates a new discovery client.
-func NewDiscoveryClient(socket string) (*DiscoveryClient, error) {
-	if socket == "" {
-		socket = DefaultSocket
-	}
-
-	client, err := containerd.New(socket)
-	if err != nil {
-		return nil, fmt.Errorf("failed to connect to containerd at %s: %w", socket, err)
-	}
-
-	return &DiscoveryClient{
-		client: client,
-		socket: socket,
-	}, nil
-}
-
-// Close closes the containerd client connection.
-func (c *DiscoveryClient) Close() error {
-	if c.client != nil {
-		return c.client.Close()
-	}
-	return nil
-}
-
-// ResolveContainer resolves a container ID to its process information.
-// This retrieves configuration from containerd RPCs (OCI spec, labels, image)
-// and runtime paths from /proc (rootfs access path).
-func (c *DiscoveryClient) ResolveContainer(ctx context.Context, containerID string) (*ContainerInfo, error) {
-	// Use the Kubernetes namespace for containerd
-	ctx = namespaces.WithNamespace(ctx, K8sNamespace)
-
-	// Load the container
-	container, err := c.client.LoadContainer(ctx, containerID)
-	if err != nil {
-		return nil, fmt.Errorf("failed to load container %s: %w", containerID, err)
-	}
-
-	// Get the task (running process)
-	task, err := container.Task(ctx, nil)
-	if err != nil {
-		return nil, fmt.Errorf("failed to get task for container %s: %w", containerID, err)
-	}
-
-	// Get the PID
-	pid := task.Pid()
-
-	// Get container image
-	image, err := container.Image(ctx)
-	if err != nil {
-		return nil, fmt.Errorf("failed to get image for container %s: %w", containerID, err)
-	}
-
-	// Get OCI spec from containerd - this contains mount config, namespace config, etc.
-	// This is preferred over parsing /proc for configuration data.
-	spec, err := container.Spec(ctx)
-	if err != nil {
-		return nil, fmt.Errorf("failed to get spec for container %s: %w", containerID, err)
-	}
-
-	// Get container labels (includes K8s pod info)
-	labels, err := container.Labels(ctx)
-	if err != nil {
-		// Labels are optional, don't fail
-		labels = make(map[string]string)
-	}
-
-	// Construct the bundle path where containerd stores the container runtime files
-	// Standard containerd layout: /run/containerd/io.containerd.runtime.v2.task/<namespace>/<container_id>/
-	containerdRunRoot := os.Getenv("CONTAINERD_RUN_ROOT")
-	if containerdRunRoot == "" {
-		containerdRunRoot = "/run/containerd"
-	}
-	bundlePath := filepath.Join(containerdRunRoot, "io.containerd.runtime.v2.task", K8sNamespace, containerID)
-
-	// Get the rootfs path from the OCI spec (usually "rootfs" relative to bundle)
-	rootfsRelPath := "rootfs"
-	if spec.Root != nil && spec.Root.Path != "" {
-		rootfsRelPath = spec.Root.Path
-	}
-
-	// Construct full rootfs path
-	var rootFS string
-	if filepath.IsAbs(rootfsRelPath) {
-		rootFS = rootfsRelPath
-	} else {
-		rootFS = filepath.Join(bundlePath, rootfsRelPath)
-	}
-
-	return &ContainerInfo{
-		ContainerID: containerID,
-		PID:         pid,
-		RootFS:      rootFS,
-		BundlePath:  bundlePath,
-		Image:       image.Name(),
-		Spec:        spec,
-		Labels:      labels,
-	}, nil
-}
-
-// GetMounts returns the mount configuration from the OCI spec.
-// This is preferred over parsing /proc/mountinfo for configuration,
-// though /proc is still needed for runtime mount state.
-func (info *ContainerInfo) GetMounts() []MountInfo {
-	if info.Spec == nil || info.Spec.Mounts == nil {
-		return nil
-	}
-
-	mounts := make([]MountInfo, len(info.Spec.Mounts))
-	for i, m := range info.Spec.Mounts {
-		mounts[i] = MountInfo{
-			Destination: m.Destination,
-			Source:      m.Source,
-			Type:        m.Type,
-			Options:     m.Options,
-		}
-	}
-	return mounts
-}
-
-// GetNamespaces returns the namespace configuration from the OCI spec.
-func (info *ContainerInfo) GetNamespaces() []NamespaceConfig {
-	if info.Spec == nil || info.Spec.Linux == nil {
-		return nil
-	}
-
-	namespaces := make([]NamespaceConfig, len(info.Spec.Linux.Namespaces))
-	for i, ns := range info.Spec.Linux.Namespaces {
-		namespaces[i] = NamespaceConfig{
-			Type: string(ns.Type),
-			Path: ns.Path,
-		}
-	}
-	return namespaces
-}
-
-// GetMaskedPaths returns the masked paths from the OCI spec.
-func (info *ContainerInfo) GetMaskedPaths() []string {
-	if info.Spec == nil || info.Spec.Linux == nil {
-		return nil
-	}
-	return info.Spec.Linux.MaskedPaths
-}
-
-// GetReadonlyPaths returns the readonly paths from the OCI spec.
-func (info *ContainerInfo) GetReadonlyPaths() []string {
-	if info.Spec == nil || info.Spec.Linux == nil {
-		return nil
-	}
-	return info.Spec.Linux.ReadonlyPaths
-}
-
-// GetRootfsPath returns the rootfs path from the OCI spec.
-// Note: For CRIU, use info.RootFS which is the /proc/<pid>/root path.
-func (info *ContainerInfo) GetRootfsPath() string {
-	if info.Spec == nil || info.Spec.Root == nil {
-		return ""
-	}
-	return info.Spec.Root.Path
-}
-
-// IsRootReadonly returns whether the root filesystem is readonly.
-func (info *ContainerInfo) IsRootReadonly() bool {
-	if info.Spec == nil || info.Spec.Root == nil {
-		return false
-	}
-	return info.Spec.Root.Readonly
-}
-
-// GetHostname returns the container's hostname from the OCI spec.
-func (info *ContainerInfo) GetHostname() string {
-	if info.Spec == nil {
-		return ""
-	}
-	return info.Spec.Hostname
-}
-
-// ListContainers lists all containers in the K8s namespace.
-func (c *DiscoveryClient) ListContainers(ctx context.Context) ([]string, error) {
-	ctx = namespaces.WithNamespace(ctx, K8sNamespace)
-
-	containers, err := c.client.Containers(ctx)
-	if err != nil {
-		return nil, fmt.Errorf("failed to list containers: %w", err)
-	}
-
-	ids := make([]string, len(containers))
-	for i, container := range containers {
-		ids[i] = container.ID()
-	}
-
-	return ids, nil
-}
-
-// GetContainerLabels returns the labels for a container.
-func (c *DiscoveryClient) GetContainerLabels(ctx context.Context, containerID string) (map[string]string, error) {
-	ctx = namespaces.WithNamespace(ctx, K8sNamespace)
-
-	container, err := c.client.LoadContainer(ctx, containerID)
-	if err != nil {
-		return nil, fmt.Errorf("failed to load container %s: %w", containerID, err)
-	}
-
-	labels, err := container.Labels(ctx)
-	if err != nil {
-		return nil, fmt.Errorf("failed to get labels for container %s: %w", containerID, err)
-	}
-
-	return labels, nil
-}

deploy/chrek/pkg/checkpoint/k8s/volumes.go

-// Package k8s provides Kubernetes-specific functionality for checkpoint operations.
-// This includes volume type discovery via K8s API and containerd container discovery.
-package k8s
-
-import (
-	"context"
-	"fmt"
-	"strings"
-
-	corev1 "k8s.io/api/core/v1"
-	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
-	"k8s.io/client-go/kubernetes"
-	"k8s.io/client-go/rest"
-	"k8s.io/client-go/tools/clientcmd"
-)
-
-// VolumeInfo contains Kubernetes volume information for a mount.
-type VolumeInfo struct {
-	VolumeName string // Name from pod.spec.volumes[].name
-	VolumeType string // Type: emptyDir, configMap, secret, persistentVolumeClaim, etc.
-	MountPath  string // Container path from volumeMounts[].mountPath
-	SubPath    string // SubPath if specified
-	ReadOnly   bool   // Whether mount is read-only
-
-	// Type-specific details
-	ConfigMapName string // For configMap volumes
-	SecretName    string // For secret volumes
-	PVCName       string // For persistentVolumeClaim volumes
-}
-
-// K8sClient wraps the Kubernetes clientset for volume discovery.
-type K8sClient struct {
-	clientset *kubernetes.Clientset
-}
-
-// NewK8sClient creates a new Kubernetes client.
-// It attempts in-cluster config first, then falls back to kubeconfig.
-func NewK8sClient() (*K8sClient, error) {
-	config, err := rest.InClusterConfig()
-	if err != nil {
-		// Fall back to kubeconfig for local development
-		config, err = clientcmd.BuildConfigFromFlags("", clientcmd.RecommendedHomeFile)
-		if err != nil {
-			return nil, fmt.Errorf("failed to create k8s config: %w", err)
-		}
-	}
-
-	clientset, err := kubernetes.NewForConfig(config)
-	if err != nil {
-		return nil, fmt.Errorf("failed to create k8s clientset: %w", err)
-	}
-
-	return &K8sClient{clientset: clientset}, nil
-}
-
-// NewK8sClientWithConfig creates a client with explicit config.
-func NewK8sClientWithConfig(config *rest.Config) (*K8sClient, error) {
-	clientset, err := kubernetes.NewForConfig(config)
-	if err != nil {
-		return nil, fmt.Errorf("failed to create k8s clientset: %w", err)
-	}
-	return &K8sClient{clientset: clientset}, nil
-}
-
-// GetPodVolumes returns volume information for all mounts in a container.
-// Returns a map from mount path to VolumeInfo.
-func (c *K8sClient) GetPodVolumes(ctx context.Context, namespace, podName, containerName string) (map[string]*VolumeInfo, error) {
-	pod, err := c.clientset.CoreV1().Pods(namespace).Get(ctx, podName, metav1.GetOptions{})
-	if err != nil {
-		return nil, fmt.Errorf("failed to get pod %s/%s: %w", namespace, podName, err)
-	}
-
-	return ExtractVolumeInfo(pod, containerName)
-}
-
-// ExtractVolumeInfo extracts volume information from a Pod spec.
-// This is the core logic that maps volumeMounts to volumes and determines types.
-func ExtractVolumeInfo(pod *corev1.Pod, containerName string) (map[string]*VolumeInfo, error) {
-	// Build volume name -> type mapping from pod.spec.volumes
-	volumeTypes := make(map[string]*volumeDetails)
-	for _, vol := range pod.Spec.Volumes {
-		volumeTypes[vol.Name] = getVolumeDetails(&vol)
-	}
-
-	// Find the target container
-	var container *corev1.Container
-	for i := range pod.Spec.Containers {
-		if pod.Spec.Containers[i].Name == containerName {
-			container = &pod.Spec.Containers[i]
-			break
-		}
-	}
-	if container == nil {
-		// Try init containers
-		for i := range pod.Spec.InitContainers {
-			if pod.Spec.InitContainers[i].Name == containerName {
-				container = &pod.Spec.InitContainers[i]
-				break
-			}
-		}
-	}
-	if container == nil {
-		return nil, fmt.Errorf("container %s not found in pod", containerName)
-	}
-
-	// Build mount path -> volume info mapping
-	result := make(map[string]*VolumeInfo)
-	for _, mount := range container.VolumeMounts {
-		details, ok := volumeTypes[mount.Name]
-		if !ok {
-			continue // Mount references unknown volume
-		}
-
-		result[mount.MountPath] = &VolumeInfo{
-			VolumeName:    mount.Name,
-			VolumeType:    details.volumeType,
-			MountPath:     mount.MountPath,
-			SubPath:       mount.SubPath,
-			ReadOnly:      mount.ReadOnly,
-			ConfigMapName: details.configMapName,
-			SecretName:    details.secretName,
-			PVCName:       details.pvcName,
-		}
-	}
-
-	return result, nil
-}
-
-// volumeDetails holds extracted volume type information.
-type volumeDetails struct {
-	volumeType    string
-	configMapName string
-	secretName    string
-	pvcName       string
-}
-
-// getVolumeDetails extracts type and details from a Volume spec.
-func getVolumeDetails(vol *corev1.Volume) *volumeDetails {
-	d := &volumeDetails{volumeType: "unknown"}
-
-	switch {
-	case vol.EmptyDir != nil:
-		d.volumeType = "emptyDir"
-	case vol.ConfigMap != nil:
-		d.volumeType = "configMap"
-		d.configMapName = vol.ConfigMap.Name
-	case vol.Secret != nil:
-		d.volumeType = "secret"
-		d.secretName = vol.Secret.SecretName
-	case vol.PersistentVolumeClaim != nil:
-		d.volumeType = "persistentVolumeClaim"
-		d.pvcName = vol.PersistentVolumeClaim.ClaimName
-	case vol.HostPath != nil:
-		d.volumeType = "hostPath"
-	case vol.Projected != nil:
-		d.volumeType = "projected"
-	case vol.DownwardAPI != nil:
-		d.volumeType = "downwardAPI"
-	case vol.CSI != nil:
-		d.volumeType = "csi"
-	case vol.NFS != nil:
-		d.volumeType = "nfs"
-	case vol.ISCSI != nil:
-		d.volumeType = "iscsi"
-	case vol.GCEPersistentDisk != nil:
-		d.volumeType = "gcePersistentDisk"
-	case vol.AWSElasticBlockStore != nil:
-		d.volumeType = "awsElasticBlockStore"
-	case vol.AzureDisk != nil:
-		d.volumeType = "azureDisk"
-	case vol.AzureFile != nil:
-		d.volumeType = "azureFile"
-	case vol.CephFS != nil:
-		d.volumeType = "cephfs"
-	case vol.Cinder != nil:
-		d.volumeType = "cinder"
-	case vol.FC != nil:
-		d.volumeType = "fc"
-	case vol.FlexVolume != nil:
-		d.volumeType = "flexVolume"
-	case vol.Flocker != nil:
-		d.volumeType = "flocker"
-	case vol.GitRepo != nil:
-		d.volumeType = "gitRepo"
-	case vol.Glusterfs != nil:
-		d.volumeType = "glusterfs"
-	case vol.PhotonPersistentDisk != nil:
-		d.volumeType = "photonPersistentDisk"
-	case vol.PortworxVolume != nil:
-		d.volumeType = "portworxVolume"
-	case vol.Quobyte != nil:
-		d.volumeType = "quobyte"
-	case vol.RBD != nil:
-		d.volumeType = "rbd"
-	case vol.ScaleIO != nil:
-		d.volumeType = "scaleIO"
-	case vol.StorageOS != nil:
-		d.volumeType = "storageos"
-	case vol.VsphereVolume != nil:
-		d.volumeType = "vsphereVolume"
-	case vol.Ephemeral != nil:
-		d.volumeType = "ephemeral"
-	}
-
-	return d
-}
-
-// DetectVolumeTypeFromPath attempts to identify volume type from kubelet path patterns.
-// This is a best-effort fallback; accurate volume types require K8s API access via GetPodVolumes.
-func DetectVolumeTypeFromPath(hostPath string) (volumeType, volumeName string) {
-	volumeType = "unknown"
-	volumeName = ""
-
-	// Map of path patterns to volume types
-	patterns := map[string]string{
-		"/kubernetes.io~empty-dir/":             "emptyDir",
-		"/kubernetes.io~configmap/":             "configMap",
-		"/kubernetes.io~secret/":                "secret",
-		"/kubernetes.io~projected/":             "projected",
-		"/kubernetes.io~downward-api/":          "downwardAPI",
-		"/kubernetes.io~persistentvolumeclaim/": "persistentVolumeClaim",
-		"/kubernetes.io~hostpath/":              "hostPath",
-	}
-
-	for pattern, vType := range patterns {
-		if strings.Contains(hostPath, pattern) {
-			volumeType = vType
-			// Extract volume name from path
-			parts := strings.Split(hostPath, pattern)
-			if len(parts) > 1 {
-				volumeName = strings.Split(parts[1], "/")[0]
-			}
-			break
-		}
-	}
-
-	return volumeType, volumeName
-}

deploy/chrek/pkg/checkpoint/metadata_builder.go

-// metadata_builder provides checkpoint metadata construction.
-package checkpoint
-
-import (
-	"context"
-	"strings"
-
-	"github.com/sirupsen/logrus"
-
-	checkpointk8s "github.com/ai-dynamo/dynamo/deploy/chrek/pkg/checkpoint/k8s"
-	"github.com/ai-dynamo/dynamo/deploy/chrek/pkg/common"
-)
-
-// MetadataBuilderConfig holds configuration for building checkpoint metadata.
-type MetadataBuilderConfig struct {
-	CheckpointID  string
-	NodeName      string
-	ContainerID   string
-	ContainerName string
-	PodName       string
-	PodNamespace  string
-	PID           int
-	CUDAPluginDir string
-}
-
-// BuildCheckpointMetadata constructs checkpoint metadata from container state.
-func BuildCheckpointMetadata(
-	ctx context.Context,
-	cfg MetadataBuilderConfig,
-	containerInfo *checkpointk8s.ContainerInfo,
-	mounts []MountMapping,
-	namespaces map[NamespaceType]*NamespaceInfo,
-	k8sClient *checkpointk8s.K8sClient,
-	log *logrus.Entry,
-) *common.CheckpointMetadata {
-	meta := common.NewCheckpointMetadata(cfg.CheckpointID)
-	meta.SourceNode = cfg.NodeName
-	meta.ContainerID = cfg.ContainerID
-	meta.PodName = cfg.PodName
-	meta.PodNamespace = cfg.PodNamespace
-	meta.PID = cfg.PID
-	meta.Image = containerInfo.Image
-
-	// Populate OCI spec derived paths
-	meta.MaskedPaths = containerInfo.GetMaskedPaths()
-	meta.ReadonlyPaths = containerInfo.GetReadonlyPaths()
-
-	// Build mount metadata
-	ociMountByDest := buildOCIMountLookup(containerInfo, meta)
-
-	// Get K8s volume types if available
-	k8sVolumes := getK8sVolumes(ctx, k8sClient, cfg, log)
-
-	// Add mount metadata
-	for _, mount := range mounts {
-		mountMeta := buildMountMetadata(mount, k8sVolumes, ociMountByDest)
-		meta.Mounts = append(meta.Mounts, mountMeta)
-	}
-
-	// Add namespace metadata
-	for nsType, nsInfo := range namespaces {
-		meta.Namespaces = append(meta.Namespaces, common.NamespaceMetadata{
-			Type:       string(nsType),
-			Inode:      nsInfo.Inode,
-			IsExternal: nsInfo.IsExternal,
-		})
-	}
-
-	// Set CRIU options (hardcoded as always-on for K8s, stored for compatibility)
-	meta.CRIUOptions = common.CRIUOptionsMetadata{
-		TcpEstablished: false, // Always false - we close TCP connections
-		TcpClose:       true,  // Always true - pod IPs change on restore
-		ShellJob:       true,  // Always true - containers are session leaders
-		FileLocks:      true,  // Always true - apps use file locks
-		LeaveRunning:   true,  // Always true - keep process running after checkpoint
-		LinkRemap:      true,  // Always true - handle deleted-but-open files
-		ExtMasters:     true,  // Always true - external bind mount masters
-	}
-
-	return meta
-}
-
-// buildOCIMountLookup builds a lookup map from OCI mounts and populates bind mount destinations.
-func buildOCIMountLookup(containerInfo *checkpointk8s.ContainerInfo, meta *common.CheckpointMetadata) map[string]checkpointk8s.MountInfo {
-	ociMounts := containerInfo.GetMounts()
-	ociMountByDest := make(map[string]checkpointk8s.MountInfo)
-	for _, m := range ociMounts {
-		ociMountByDest[m.Destination] = m
-		if m.Type == "bind" {
-			meta.BindMountDests = append(meta.BindMountDests, m.Destination)
-		}
-	}
-	return ociMountByDest
-}
-
-// getK8sVolumes fetches volume types from K8s API if available.
-func getK8sVolumes(ctx context.Context, k8sClient *checkpointk8s.K8sClient, cfg MetadataBuilderConfig, log *logrus.Entry) map[string]*checkpointk8s.VolumeInfo {
-	if k8sClient == nil || cfg.PodNamespace == "" || cfg.PodName == "" || cfg.ContainerName == "" {
-		return nil
-	}
-
-	k8sVolumes, err := k8sClient.GetPodVolumes(ctx, cfg.PodNamespace, cfg.PodName, cfg.ContainerName)
-	if err != nil {
-		log.WithError(err).Warn("Failed to get volume types from K8s API, falling back to path-based detection")
-		return nil
-	}
-	log.WithField("volume_count", len(k8sVolumes)).Debug("Got volume types from K8s API")
-	return k8sVolumes
-}
-
-// buildMountMetadata constructs metadata for a single mount.
-func buildMountMetadata(mount MountMapping, k8sVolumes map[string]*checkpointk8s.VolumeInfo, ociMountByDest map[string]checkpointk8s.MountInfo) common.MountMetadata {
-	var volumeType, volumeName string
-
-	// Try K8s API first for accurate volume types
-	if k8sVolumes != nil {
-		if volInfo, ok := k8sVolumes[mount.InsidePath]; ok {
-			volumeType = volInfo.VolumeType
-			volumeName = volInfo.VolumeName
-		}
-	}
-
-	// Fall back to path-based detection if K8s API didn't provide info
-	if volumeType == "" {
-		volumeType, volumeName = checkpointk8s.DetectVolumeTypeFromPath(mount.OutsidePath)
-	}
-
-	mountMeta := common.MountMetadata{
-		ContainerPath: mount.InsidePath,
-		HostPath:      mount.OutsidePath,
-		VolumeType:    volumeType,
-		VolumeName:    volumeName,
-		FSType:        mount.FSType,
-		ReadOnly:      strings.Contains(mount.Options, "ro"),
-	}
-
-	// Cross-reference with OCI spec mount if available
-	if ociMount, ok := ociMountByDest[mount.InsidePath]; ok {
-		mountMeta.OCISource = ociMount.Source
-		mountMeta.OCIType = ociMount.Type
-		mountMeta.OCIOptions = ociMount.Options
-	}
-
-	return mountMeta
-}
-

deploy/chrek/pkg/checkpoint/mounts.go

-// mounts provides mount parsing from /proc for CRIU checkpoint.
-// This is used for runtime mount state that requires /proc inspection.
+// mounts parses runtime mount state from /proc.
 package checkpoint
 
 import (
-	"bufio"
 	"fmt"
-	"os"
+	"path"
+	"path/filepath"
 	"strings"
-)
 
-// MountMapping represents an external mount for CRIU
-type MountMapping struct {
-	InsidePath  string // Path inside container (mount point)
-	OutsidePath string // Path on host (source)
-	FSType      string // Filesystem type
-	Source      string // Mount source
-	Options     string // Mount options
-}
+	specs "github.com/opencontainers/runtime-spec/specs-go"
 
-// System mount types that should be filtered out
-var systemMountTypes = map[string]bool{
-	"proc":        true,
-	"sysfs":       true,
-	"devpts":      true,
-	"mqueue":      true,
-	"tmpfs":       true, // Note: some tmpfs mounts may need special handling
-	"cgroup":      true,
-	"cgroup2":     true,
-	"securityfs":  true,
-	"debugfs":     true,
-	"tracefs":     true,
-	"fusectl":     true,
-	"configfs":    true,
-	"devtmpfs":    true,
-	"hugetlbfs":   true,
-	"pstore":      true,
-	"bpf":         true,
-}
+	"github.com/ai-dynamo/dynamo/deploy/chrek/pkg/common"
+)
 
-// System mount paths that should always be filtered
-var systemMountPaths = map[string]bool{
-	"/proc":        true,
-	"/sys":         true,
-	"/dev":         true,
-	"/dev/pts":     true,
-	"/dev/shm":     true,
-	"/dev/mqueue":  true,
-	"/run":         true,
-	"/run/secrets": true,
+type MountInfo struct {
+	MountID      string
+	ParentID     string
+	MountPoint   string
+	Root         string
+	FSType       string
+	Source       string
+	Options      string
+	SuperOptions string
 }
 
-// ParseMountInfo parses /proc/<pid>/mountinfo and returns bind mounts
-// that need to be handled by CRIU as external mounts
-func ParseMountInfo(pid int, hostProc string) ([]MountMapping, error) {
-	if hostProc == "" {
-		hostProc = "/proc"
-	}
-
-	mountinfoPath := fmt.Sprintf("%s/%d/mountinfo", hostProc, pid)
-	file, err := os.Open(mountinfoPath)
-	if err != nil {
-		return nil, fmt.Errorf("failed to open mountinfo: %w", err)
-	}
-	defer file.Close()
-
-	var mounts []MountMapping
-	scanner := bufio.NewScanner(file)
-
-	for scanner.Scan() {
-		line := scanner.Text()
-		mount, skip := parseMountInfoLine(line)
-		if skip {
-			continue
-		}
-		mounts = append(mounts, mount)
-	}
-
-	if err := scanner.Err(); err != nil {
-		return nil, fmt.Errorf("error reading mountinfo: %w", err)
-	}
-
-	return mounts, nil
+// MountPolicy is the classified mount plan for CRIU dump options.
+type MountPolicy struct {
+	Externalized []string
+	Skipped      []string
 }
 
-// parseMountInfoLine parses a single line from mountinfo
-// Returns the mount mapping and whether to skip this mount
+// BuildMountPolicy classifies mounts into CRIU extMnt and skipMnt lists.
 //
-// mountinfo format:
-// 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue
-// (1)(2)(3)   (4)   (5)      (6)     (7)   (8) (9)   (10)         (11)
+// Rule order and precedence (top to bottom):
+//  1. Skip non-OCI proc/sys submounts and non-OCI runtime /run submounts.
+//     These mounts are typically node/kernel/runtime specific and are the
+//     highest-risk source of cross-node restore failures, so skip wins.
+//  2. Externalize mounts owned by runtime/OCI:
+//     - "/" (rootfs is recreated by runtime in OCI restore path)
+//     - OCI mount destinations
+//     - OCI masked/readonly paths
+//  3. Externalize non-OCI bind-like mounts (mount root is not "/" or ".").
+//     This captures runtime-injected file mounts (for example driver files)
+//     so CRIU does not try to recreate them from checkpoint data.
+//  4. Anything else is left unflagged and handled by CRIU default behavior.
 //
-// (1) mount ID
-// (2) parent ID
-// (3) major:minor
-// (4) root: root of the mount within the filesystem (host-side path for bind mounts)
-// (5) mount point: mount point relative to process's root
-// (6) mount options
-// (7) optional fields (terminated by single hyphen)
-// (8) separator (hyphen)
-// (9) filesystem type
-// (10) mount source (device)
-// (11) super options
-func parseMountInfoLine(line string) (MountMapping, bool) {
-	fields := strings.Fields(line)
-	if len(fields) < 10 {
-		return MountMapping{}, true
-	}
+// Precedence: skip > externalize. If a path is classified as skipped, it is
+// removed from the externalized set.
+func BuildMountPolicy(mountInfo []MountInfo, ociSpec *specs.Spec, rootFS string) *MountPolicy {
+	ociManagedSet := collectOCIManagedDestinations(ociSpec, rootFS)
 
-	root := fields[3]       // Host-side path within the filesystem (important for bind mounts)
-	mountPoint := fields[4] // Container-side mount point
-	mountOptions := fields[5]
+	externalizedSet := make(map[string]struct{}, len(mountInfo)+len(ociManagedSet))
+	skippedSet := make(map[string]struct{}, len(mountInfo))
 
-	// Find separator (-) to get fstype and source
-	sepIdx := -1
-	for i, f := range fields {
-		if f == "-" {
-			sepIdx = i
-			break
+	for _, mount := range mountInfo {
+		mp := normalizeMountPath(mount.MountPoint)
+		if mp == "" {
+			continue
 		}
-	}
 
-	if sepIdx == -1 || sepIdx+2 >= len(fields) {
-		return MountMapping{}, true
-	}
+		source := path.Clean(strings.TrimSpace(mount.Source))
+		root := path.Clean(strings.TrimSpace(mount.Root))
+		isOCIManaged := false
+		if _, ok := ociManagedSet[mp]; ok {
+			isOCIManaged = true
+		}
+		if !isOCIManaged && strings.HasPrefix(mp, "/run/") {
+			if _, ok := ociManagedSet["/var"+mp]; ok {
+				isOCIManaged = true
+			}
+		}
+		if !isOCIManaged && strings.HasPrefix(mp, "/var/run/") {
+			if _, ok := ociManagedSet[strings.TrimPrefix(mp, "/var")]; ok {
+				isOCIManaged = true
+			}
+		}
 
-	fsType := fields[sepIdx+1]
-	source := fields[sepIdx+2]
-	superOptions := ""
-	if sepIdx+3 < len(fields) {
-		superOptions = fields[sepIdx+3]
-	}
+		// Runtime-owned /run mounts are usually ephemeral tmpfs/overlay mounts
+		// or bind-like mounts sourced from host runtime directories.
+		// We skip these unless OCI explicitly manages that destination.
+		isRunRuntimeMount := strings.HasPrefix(mp, "/run/") &&
+			(mount.FSType == "tmpfs" ||
+				mount.FSType == "overlay" ||
+				strings.HasPrefix(source, "/run/") ||
+				strings.HasPrefix(source, "/var/run/") ||
+				strings.HasPrefix(root, "/run/") ||
+				strings.HasPrefix(root, "/var/run/"))
+
+		if !isOCIManaged && (strings.HasPrefix(mp, "/proc/") || strings.HasPrefix(mp, "/sys/") || isRunRuntimeMount) {
+			skippedSet[mp] = struct{}{}
+			delete(externalizedSet, mp)
+			continue
+		}
 
-	// Skip system mount types
-	if systemMountTypes[fsType] {
-		return MountMapping{}, true
+		if mp == "/" || isOCIManaged || (root != "." && root != "/") {
+			externalizedSet[mp] = struct{}{}
+			continue
+		}
 	}
 
-	// Skip system mount paths
-	if systemMountPaths[mountPoint] {
-		return MountMapping{}, true
+	// Ensure OCI-managed destinations are externalized, even when mountinfo does not
+	// include a direct entry (e.g., runtime-managed masked/readonly paths).
+	for mp := range ociManagedSet {
+		if _, skipped := skippedSet[mp]; skipped {
+			continue
+		}
+		externalizedSet[mp] = struct{}{}
 	}
 
-	// Skip /sys and /proc prefixed paths
-	if strings.HasPrefix(mountPoint, "/sys/") || strings.HasPrefix(mountPoint, "/proc/") {
-		return MountMapping{}, true
+	externalized := make([]string, 0, len(externalizedSet))
+	for mp := range externalizedSet {
+		externalized = append(externalized, mp)
 	}
-
-	// Skip overlay (the root filesystem itself)
-	if fsType == "overlay" && mountPoint == "/" {
-		return MountMapping{}, true
+	skipped := make([]string, 0, len(skippedSet))
+	for mp := range skippedSet {
+		skipped = append(skipped, mp)
 	}
 
-	// For bind mounts, the root field contains the actual host path
-	// Use root as OutsidePath since it gives us the host-side path for volume mounts
-	outsidePath := root
-	if root == "/" {
-		// If root is /, this isn't a bind mount from a subdirectory
-		outsidePath = source
+	return &MountPolicy{
+		Externalized: externalized,
+		Skipped:      skipped,
 	}
-
-	return MountMapping{
-		InsidePath:  mountPoint,
-		OutsidePath: outsidePath,
-		FSType:      fsType,
-		Source:      source,
-		Options:     mountOptions + "," + superOptions,
-	}, false
 }
 
-// GetBindMounts returns only bind mounts (type "bind" or with bind option)
-func GetBindMounts(pid int, hostProc string) ([]MountMapping, error) {
-	mounts, err := ParseMountInfo(pid, hostProc)
-	if err != nil {
-		return nil, err
+// collectOCIManagedDestinations returns the canonical set of OCI-owned mount
+// targets. This includes regular OCI mounts plus Linux masked/readonly paths.
+// Those masked/readonly paths may not appear as direct mountinfo entries, but
+// still need to be treated as runtime-owned and externalized.
+func collectOCIManagedDestinations(ociSpec *specs.Spec, rootFS string) map[string]struct{} {
+	set := map[string]struct{}{}
+	if ociSpec == nil {
+		return set
 	}
 
-	var bindMounts []MountMapping
-	for _, m := range mounts {
-		// Bind mounts typically show the underlying filesystem type
-		// and have paths that look like kubelet volume paths
-		if strings.Contains(m.OutsidePath, "/var/lib/kubelet/pods/") ||
-			strings.Contains(m.OutsidePath, "/volumes/") ||
-			strings.Contains(m.Options, "bind") {
-			bindMounts = append(bindMounts, m)
+	paths := make([]string, 0, len(ociSpec.Mounts))
+	for _, mount := range ociSpec.Mounts {
+		paths = append(paths, mount.Destination)
+	}
+	if ociSpec.Linux != nil {
+		paths = append(paths, ociSpec.Linux.MaskedPaths...)
+		paths = append(paths, ociSpec.Linux.ReadonlyPaths...)
+	}
+	for _, raw := range paths {
+		if p := normalizeOCIDestinationPath(raw, rootFS); p != "" {
+			set[p] = struct{}{}
 		}
 	}
 
-	return bindMounts, nil
+	return set
 }
 
-// GetKubernetesVolumeMounts returns mounts that appear to be Kubernetes volumes
-func GetKubernetesVolumeMounts(pid int, hostProc string) ([]MountMapping, error) {
-	mounts, err := ParseMountInfo(pid, hostProc)
-	if err != nil {
-		return nil, err
+// normalizeMountPath applies lexical normalization only.
+// Mountinfo paths are already kernel truth for the container namespace.
+func normalizeMountPath(raw string) string {
+	raw = strings.TrimSpace(raw)
+	if raw == "" {
+		return ""
 	}
 
-	var k8sMounts []MountMapping
-	for _, m := range mounts {
-		// Kubernetes volumes are identified by:
-		// 1. Standard kubelet paths: /var/lib/kubelet/pods/
-		// 2. Minikube/Docker paths: /var/lib/docker/volumes/minikube/_data/lib/kubelet/pods/
-		// 3. Kubernetes volume markers: kubernetes.io~empty-dir, kubernetes.io~configmap, etc.
-		if strings.Contains(m.OutsidePath, "/kubelet/pods/") ||
-			strings.Contains(m.OutsidePath, "/kubernetes.io~") ||
-			strings.Contains(m.OutsidePath, "/containerd/io.containerd") {
-			k8sMounts = append(k8sMounts, m)
-		}
+	p := path.Clean(raw)
+	if !strings.HasPrefix(p, "/") {
+		p = "/" + p
 	}
-
-	return k8sMounts, nil
+	return path.Clean(p)
 }
 
-// AllMountInfo represents a mount entry from /proc/<pid>/mountinfo
-// This includes ALL mounts without filtering, which CRIU captures during checkpoint.
-type AllMountInfo struct {
-	MountID      string // Mount ID
-	ParentID     string // Parent mount ID
-	MountPoint   string // Mount point inside container (container-side path)
-	Root         string // Root of mount within filesystem (host-side path for bind mounts)
-	FSType       string // Filesystem type
-	Source       string // Mount source
-	Options      string // Mount options
-	SuperOptions string // Super block options
-}
-
-// GetAllMountsFromMountinfo parses /proc/<pid>/mountinfo and returns ALL mounts.
-// This is used for CRIU checkpoint to mark ALL mounts as external, since CRIU
-// captures everything from mountinfo, not just the filtered subset.
-// Without marking ALL mounts as external, CRIU restore fails with
-// "No mapping for <mount_id>:(null) mountpoint" errors.
-func GetAllMountsFromMountinfo(pid int, hostProc string) ([]AllMountInfo, error) {
-	if hostProc == "" {
-		hostProc = "/proc"
+// normalizeOCIDestinationPath canonicalizes OCI destinations against container
+// rootfs symlinks (for example /var/run -> /run) with lexical fallback.
+func normalizeOCIDestinationPath(raw, rootFS string) string {
+	p := normalizeMountPath(raw)
+	if p == "" || rootFS == "" {
+		return p
 	}
 
-	mountinfoPath := fmt.Sprintf("%s/%d/mountinfo", hostProc, pid)
-	file, err := os.Open(mountinfoPath)
+	hostPath := filepath.Join(rootFS, strings.TrimPrefix(p, "/"))
+	resolved, err := filepath.EvalSymlinks(hostPath)
 	if err != nil {
-		return nil, fmt.Errorf("failed to open mountinfo: %w", err)
+		return p
 	}
-	defer file.Close()
-
-	var mounts []AllMountInfo
-	scanner := bufio.NewScanner(file)
 
-	for scanner.Scan() {
-		line := scanner.Text()
-		mount, err := parseAllMountInfoLine(line)
-		if err != nil {
-			continue // Skip malformed lines
-		}
-		mounts = append(mounts, mount)
+	rel, err := filepath.Rel(rootFS, resolved)
+	if err != nil {
+		return p
 	}
-
-	if err := scanner.Err(); err != nil {
-		return nil, fmt.Errorf("error reading mountinfo: %w", err)
+	rel = filepath.ToSlash(rel)
+	if rel == "." {
+		return "/"
 	}
-
-	return mounts, nil
-}
-
-// parseAllMountInfoLine parses a single line from mountinfo without filtering.
-// mountinfo format:
-// 36 35 98:0 /mnt1 /mnt2 rw,noatime master:1 - ext3 /dev/root rw,errors=continue
-// (1)(2)(3)   (4)   (5)      (6)     (7)   (8) (9)   (10)         (11)
-func parseAllMountInfoLine(line string) (AllMountInfo, error) {
-	fields := strings.Fields(line)
-	if len(fields) < 10 {
-		return AllMountInfo{}, fmt.Errorf("malformed mountinfo line: %s", line)
+	if strings.HasPrefix(rel, "../") || rel == ".." {
+		return p
 	}
 
-	mountID := fields[0]
-	parentID := fields[1]
-	root := fields[3]       // Host-side path within the filesystem
-	mountPoint := fields[4] // Container-side mount point
-	mountOptions := fields[5]
-
-	// Find separator (-) to get fstype and source
-	sepIdx := -1
-	for i, f := range fields {
-		if f == "-" {
-			sepIdx = i
-			break
-		}
-	}
+	return normalizeMountPath("/" + rel)
+}
 
-	if sepIdx == -1 || sepIdx+2 >= len(fields) {
-		return AllMountInfo{}, fmt.Errorf("malformed mountinfo line (no separator): %s", line)
+func ReadMountInfoFromHostProcPath(pid int) ([]MountInfo, error) {
+	mountinfoPath := fmt.Sprintf("%s/%d/mountinfo", HostProcPath, pid)
+	parsedMounts, err := common.ParseMountInfoFile(mountinfoPath)
+	if err != nil {
+		return nil, fmt.Errorf("failed to parse mountinfo at %s: %w", mountinfoPath, err)
 	}
 
-	fsType := fields[sepIdx+1]
-	source := fields[sepIdx+2]
-	superOptions := ""
-	if sepIdx+3 < len(fields) {
-		superOptions = fields[sepIdx+3]
+	mounts := make([]MountInfo, 0, len(parsedMounts))
+	for _, parsed := range parsedMounts {
+		mounts = append(mounts, MountInfo{
+			MountID:      parsed.MountID,
+			ParentID:     parsed.ParentID,
+			MountPoint:   parsed.Path,
+			Root:         parsed.Root,
+			FSType:       parsed.FSType,
+			Source:       parsed.Source,
+			Options:      parsed.Options,
+			SuperOptions: parsed.SuperOpts,
+		})
 	}
 
-	return AllMountInfo{
-		MountID:      mountID,
-		ParentID:     parentID,
-		MountPoint:   mountPoint,
-		Root:         root,
-		FSType:       fsType,
-		Source:       source,
-		Options:      mountOptions,
-		SuperOptions: superOptions,
-	}, nil
+	return mounts, nil
 }

deploy/chrek/pkg/checkpoint/namespaces.go

@@ -3,12 +3,17 @@ package checkpoint
 
 import (
 	"fmt"
-	"os"
-	"strings"
 
 	"golang.org/x/sys/unix"
 )
 
+// NamespaceManifestEntry stores namespace information saved in checkpoint manifests.
+type NamespaceManifestEntry struct {
+	Type       string `yaml:"type"`       // net, pid, mnt, etc.
+	Inode      uint64 `yaml:"inode"`      // Namespace inode
+	IsExternal bool   `yaml:"isExternal"` // Whether namespace is external (shared)
+}
+
 // NamespaceType represents a Linux namespace type
 type NamespaceType string
 
@@ -26,17 +31,29 @@ const (
 type NamespaceInfo struct {
 	Type       NamespaceType
 	Inode      uint64
-	Path       string
 	IsExternal bool // Whether NS is external (shared with pause container)
 }
 
-// GetNamespaceInode returns the inode number for a namespace
-func GetNamespaceInode(pid int, nsType NamespaceType, hostProc string) (uint64, error) {
-	if hostProc == "" {
-		hostProc = "/proc"
+// NewNamespaceManifestEntries constructs namespace manifest entries from introspected namespaces.
+func NewNamespaceManifestEntries(namespaces map[NamespaceType]*NamespaceInfo) []NamespaceManifestEntry {
+	if len(namespaces) == 0 {
+		return nil
 	}
 
-	nsPath := fmt.Sprintf("%s/%d/ns/%s", hostProc, pid, nsType)
+	result := make([]NamespaceManifestEntry, 0, len(namespaces))
+	for nsType, nsInfo := range namespaces {
+		result = append(result, NamespaceManifestEntry{
+			Type:       string(nsType),
+			Inode:      nsInfo.Inode,
+			IsExternal: nsInfo.IsExternal,
+		})
+	}
+	return result
+}
+
+// GetNamespaceInode returns the inode number for a namespace
+func GetNamespaceInode(pid int, nsType NamespaceType) (uint64, error) {
+	nsPath := fmt.Sprintf("%s/%d/ns/%s", HostProcPath, pid, nsType)
 	var stat unix.Stat_t
 	if err := unix.Stat(nsPath, &stat); err != nil {
 		return 0, fmt.Errorf("failed to stat namespace %s: %w", nsPath, err)
@@ -46,12 +63,8 @@ func GetNamespaceInode(pid int, nsType NamespaceType, hostProc string) (uint64,
 }
 
 // GetNamespaceInfo returns detailed namespace information
-func GetNamespaceInfo(pid int, nsType NamespaceType, hostProc string) (*NamespaceInfo, error) {
-	if hostProc == "" {
-		hostProc = "/proc"
-	}
-
-	nsPath := fmt.Sprintf("%s/%d/ns/%s", hostProc, pid, nsType)
+func GetNamespaceInfo(pid int, nsType NamespaceType) (*NamespaceInfo, error) {
+	nsPath := fmt.Sprintf("%s/%d/ns/%s", HostProcPath, pid, nsType)
 
 	// Get inode
 	var stat unix.Stat_t
@@ -59,14 +72,8 @@ func GetNamespaceInfo(pid int, nsType NamespaceType, hostProc string) (*Namespac
 		return nil, fmt.Errorf("failed to stat namespace %s: %w", nsPath, err)
 	}
 
-	// Read the symlink to get the namespace identifier
-	link, err := os.Readlink(nsPath)
-	if err != nil {
-		return nil, fmt.Errorf("failed to readlink %s: %w", nsPath, err)
-	}
-
 	// Check if this is different from init's namespace (PID 1)
-	initNsPath := fmt.Sprintf("%s/1/ns/%s", hostProc, nsType)
+	initNsPath := fmt.Sprintf("%s/1/ns/%s", HostProcPath, nsType)
 	var initStat unix.Stat_t
 	isExternal := false
 	if err := unix.Stat(initNsPath, &initStat); err == nil {
@@ -77,13 +84,12 @@ func GetNamespaceInfo(pid int, nsType NamespaceType, hostProc string) (*Namespac
 	return &NamespaceInfo{
 		Type:       nsType,
 		Inode:      stat.Ino,
-		Path:       link,
 		IsExternal: isExternal,
 	}, nil
 }
 
 // GetAllNamespaces returns information about all namespaces for a process
-func GetAllNamespaces(pid int, hostProc string) (map[NamespaceType]*NamespaceInfo, error) {
+func GetAllNamespaces(pid int) (map[NamespaceType]*NamespaceInfo, error) {
 	nsTypes := []NamespaceType{
 		NamespaceNet,
 		NamespacePID,
@@ -96,66 +102,10 @@ func GetAllNamespaces(pid int, hostProc string) (map[NamespaceType]*NamespaceInf
 
 	namespaces := make(map[NamespaceType]*NamespaceInfo)
 	for _, nsType := range nsTypes {
-		info, err := GetNamespaceInfo(pid, nsType, hostProc)
-		if err != nil {
-			// Some namespaces might not exist, skip them
-			continue
+		if info, err := GetNamespaceInfo(pid, nsType); err == nil {
+			namespaces[nsType] = info
 		}
-		namespaces[nsType] = info
 	}
 
 	return namespaces, nil
 }
-
-// IsNetNamespaceExternal checks if the network namespace is external
-// (i.e., shared with the pause container in Kubernetes)
-func IsNetNamespaceExternal(pid int, hostProc string) (bool, uint64, error) {
-	info, err := GetNamespaceInfo(pid, NamespaceNet, hostProc)
-	if err != nil {
-		return false, 0, err
-	}
-	return info.IsExternal, info.Inode, nil
-}
-
-// IsPIDNamespaceExternal checks if the PID namespace is external
-func IsPIDNamespaceExternal(pid int, hostProc string) (bool, uint64, error) {
-	info, err := GetNamespaceInfo(pid, NamespacePID, hostProc)
-	if err != nil {
-		return false, 0, err
-	}
-	return info.IsExternal, info.Inode, nil
-}
-
-// OpenNamespaceFD opens a file descriptor to a namespace
-// The caller is responsible for closing the returned file
-func OpenNamespaceFD(pid int, nsType NamespaceType, hostProc string) (*os.File, error) {
-	if hostProc == "" {
-		hostProc = "/proc"
-	}
-
-	nsPath := fmt.Sprintf("%s/%d/ns/%s", hostProc, pid, nsType)
-	return os.Open(nsPath)
-}
-
-// FormatExternalNamespace formats namespace info for CRIU's External option
-// Format: <type>[<inode>]:<key>
-func FormatExternalNamespace(nsType NamespaceType, inode uint64) string {
-	key := formatNamespaceKey(nsType)
-	return fmt.Sprintf("%s[%d]:%s", nsType, inode, key)
-}
-
-// formatNamespaceKey creates the CRIU key for external namespaces
-// Format: extRoot<Type>NS (e.g., extRootNetNS, extRootPidNS)
-func formatNamespaceKey(nsType NamespaceType) string {
-	// Capitalize first letter of namespace type
-	nsName := string(nsType)
-	if len(nsName) > 0 {
-		nsName = strings.ToUpper(nsName[:1]) + nsName[1:]
-	}
-	return "extRoot" + nsName + "NS"
-}
-
-// GetNamespaceKey returns the CRIU key for a namespace type
-func GetNamespaceKey(nsType NamespaceType) string {
-	return formatNamespaceKey(nsType)
-}

deploy/chrek/pkg/checkpoint/shm.go

+// Package checkpoint provides CRIU checkpoint (dump) operations.
+package checkpoint
+
+import (
+	"fmt"
+	"io"
+	"os"
+	"path/filepath"
+
+	"github.com/sirupsen/logrus"
+)
+
+// CaptureDevShm captures files from /dev/shm to the checkpoint directory.
+// This is needed because /dev/shm is a tmpfs mount that is not part of the
+// container's overlay filesystem, so rootfs diff doesn't capture it.
+//
+// Semaphores (sem.* files) are included so that sem_unlink() calls succeed
+// after restore. The semaphore kernel state won't be perfectly restored,
+// but the files will exist for cleanup operations.
+//
+// The files are saved to <checkpointDir>/dev-shm/ and can be restored
+// using RestoreDevShm before CRIU restore.
+func CaptureDevShm(pid int, checkpointDir string, log *logrus.Entry) error {
+	// Access container's /dev/shm via /proc/<pid>/root
+	shmPath := filepath.Join(HostProcPath, fmt.Sprintf("%d/root/dev/shm", pid))
+
+	entries, err := os.ReadDir(shmPath)
+	if err != nil {
+		if os.IsNotExist(err) {
+			log.Debug("Container /dev/shm does not exist, skipping capture")
+			return nil
+		}
+		return fmt.Errorf("failed to read container /dev/shm: %w", err)
+	}
+
+	// Filter out directories
+	var filesToCapture []os.DirEntry
+	for _, entry := range entries {
+		// Skip directories (unlikely in /dev/shm but be safe)
+		if entry.IsDir() {
+			log.WithField("dir", entry.Name()).Debug("Skipping directory in /dev/shm")
+			continue
+		}
+
+		filesToCapture = append(filesToCapture, entry)
+	}
+
+	if len(filesToCapture) == 0 {
+		log.Debug("No files to capture from /dev/shm")
+		return nil
+	}
+
+	// Create destination directory
+	destDir := filepath.Join(checkpointDir, DevShmDirName)
+	if err := os.MkdirAll(destDir, 0755); err != nil {
+		return fmt.Errorf("failed to create dev-shm directory: %w", err)
+	}
+
+	var captured []string
+	var totalSize int64
+
+	for _, entry := range filesToCapture {
+		name := entry.Name()
+		srcPath := filepath.Join(shmPath, name)
+		destPath := filepath.Join(destDir, name)
+
+		info, err := entry.Info()
+		if err != nil {
+			log.WithError(err).WithField("file", name).Warn("Failed to get file info, skipping")
+			continue
+		}
+
+		size := info.Size()
+
+		// Copy the file
+		if err := copyFile(srcPath, destPath, info.Mode()); err != nil {
+			log.WithError(err).WithField("file", name).Warn("Failed to copy file, skipping")
+			continue
+		}
+
+		captured = append(captured, name)
+		totalSize += size
+
+		log.WithFields(logrus.Fields{
+			"file": name,
+			"size": size,
+		}).Debug("Captured /dev/shm file")
+	}
+
+	if len(captured) > 0 {
+		log.WithFields(logrus.Fields{
+			"count":      len(captured),
+			"total_size": totalSize,
+			"files":      captured,
+		}).Info("Captured /dev/shm files")
+	}
+
+	return nil
+}
+
+// copyFile copies a file from src to dest with the given permissions.
+func copyFile(src, dest string, mode os.FileMode) error {
+	srcFile, err := os.Open(src)
+	if err != nil {
+		return fmt.Errorf("failed to open source: %w", err)
+	}
+	defer srcFile.Close()
+
+	destFile, err := os.OpenFile(dest, os.O_CREATE|os.O_WRONLY|os.O_TRUNC, mode)
+	if err != nil {
+		return fmt.Errorf("failed to create destination: %w", err)
+	}
+	defer destFile.Close()
+
+	if _, err := io.Copy(destFile, srcFile); err != nil {
+		return fmt.Errorf("failed to copy contents: %w", err)
+	}
+
+	// Sync to ensure durability for checkpoint data
+	if err := destFile.Sync(); err != nil {
+		return fmt.Errorf("failed to sync destination: %w", err)
+	}
+
+	return nil
+}

deploy/chrek/pkg/checkpoint/storage.go

+// storage.go provides checkpoint storage I/O: write/read manifests, listing, deletion.
+package checkpoint
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"strings"
+
+	"gopkg.in/yaml.v3"
+)
+
+// WriteCheckpointManifest writes a checkpoint manifest file in the checkpoint directory.
+func WriteCheckpointManifest(checkpointDir string, data *CheckpointManifest) error {
+	content, err := yaml.Marshal(data)
+	if err != nil {
+		return fmt.Errorf("failed to marshal checkpoint manifest: %w", err)
+	}
+
+	manifestPath := filepath.Join(checkpointDir, CheckpointManifestFilename)
+	if err := os.WriteFile(manifestPath, content, 0600); err != nil {
+		return fmt.Errorf("failed to write checkpoint manifest: %w", err)
+	}
+
+	return nil
+}
+
+// ReadCheckpointManifest reads checkpoint manifest from a checkpoint directory.
+func ReadCheckpointManifest(checkpointDir string) (*CheckpointManifest, error) {
+	manifestPath := filepath.Join(checkpointDir, CheckpointManifestFilename)
+
+	content, err := os.ReadFile(manifestPath)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read checkpoint manifest: %w", err)
+	}
+
+	var data CheckpointManifest
+	if err := yaml.Unmarshal(content, &data); err != nil {
+		return nil, fmt.Errorf("failed to unmarshal checkpoint manifest: %w", err)
+	}
+
+	return &data, nil
+}
+
+// SaveDescriptors writes file descriptor information to the checkpoint directory.
+func SaveDescriptors(checkpointDir string, descriptors []string) error {
+	content, err := yaml.Marshal(descriptors)
+	if err != nil {
+		return fmt.Errorf("failed to marshal descriptors: %w", err)
+	}
+
+	descriptorsPath := filepath.Join(checkpointDir, DescriptorsFilename)
+	if err := os.WriteFile(descriptorsPath, content, 0600); err != nil {
+		return fmt.Errorf("failed to write descriptors file: %w", err)
+	}
+
+	return nil
+}
+
+// LoadDescriptors reads file descriptor information from checkpoint directory.
+func LoadDescriptors(checkpointDir string) ([]string, error) {
+	descriptorsPath := filepath.Join(checkpointDir, DescriptorsFilename)
+
+	content, err := os.ReadFile(descriptorsPath)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read descriptors file: %w", err)
+	}
+
+	var descriptors []string
+	if err := yaml.Unmarshal(content, &descriptors); err != nil {
+		return nil, fmt.Errorf("failed to unmarshal descriptors: %w", err)
+	}
+
+	return descriptors, nil
+}
+
+// ListCheckpoints returns all checkpoint IDs in the base directory.
+func ListCheckpoints(baseDir string) ([]string, error) {
+	entries, err := os.ReadDir(baseDir)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read checkpoint directory: %w", err)
+	}
+
+	var checkpoints []string
+	for _, entry := range entries {
+		if !entry.IsDir() {
+			continue
+		}
+
+		// Check if manifest file exists.
+		manifestPath := filepath.Join(baseDir, entry.Name(), CheckpointManifestFilename)
+		if _, err := os.Stat(manifestPath); err == nil {
+			checkpoints = append(checkpoints, entry.Name())
+		}
+	}
+
+	return checkpoints, nil
+}
+
+// DeleteCheckpoint removes a checkpoint directory.
+func DeleteCheckpoint(baseDir, checkpointID string) error {
+	checkpointDir := filepath.Join(baseDir, checkpointID)
+	// Ensure resolved path is within baseDir to prevent path traversal
+	absBase, _ := filepath.Abs(baseDir)
+	absDir, _ := filepath.Abs(checkpointDir)
+	if !strings.HasPrefix(absDir, absBase+string(filepath.Separator)) && absDir != absBase {
+		return fmt.Errorf("invalid checkpoint ID: resolved path %s is outside base directory %s", absDir, absBase)
+	}
+	return os.RemoveAll(checkpointDir)
+}