feat(T1.5): implement workflow pause/resume with state snapshots
- Add internal/pause package for pause/resume orchestration - Implement WorkflowSnapshot for complete state serialization - Implement SnapshotManager for snapshot storage and recovery - Implement PauseHandler for pause/resume signal handling - Implement PauseSignal and ResumeSignal types - Implement PauseState for tracking pause status Snapshot Features: - Capture complete workflow state (tasks, metrics, config) - Persist to JSON files for recovery after pod restart - Track paused_at and resumed_at timestamps - Support snapshot cleanup and batch removal - Load/save from disk with persistence layer Pause Handling: - Accept pause signals with reason and grace period - Save current state before pausing - Block workflow execution during pause - Support multiple concurrent paused workflows - Channel-based signal reception (Temporal-compatible) Resume Handling: - Accept resume signals with reason - Restore workflow state from snapshots - Continue execution from exact pause point - Update timestamps on resumption - Enable recovery after pod restarts Signal Management: - Non-blocking signal reception with timeout - WaitForPauseOrResume() for blocking operations - ConfigurableWait duration - Error handling for invalid transitions Analytics: - GetPauseStats() for pause/resume metrics - GetSnapshotStats() for snapshot inventory - Timestamp tracking (paused, resumed) - Multi-workflow state aggregation Test Coverage: - 16 snapshot tests (creation, persistence, cleanup) - 18 handler tests (signals, state, snapshots) - 34 total pause/resume tests, all passing - Edge cases: concurrent workflows, nil signals, timeouts - State transition verification Key Design: - Separate Snapshot Manager (storage) and Pause Handler (orchestration) - JSON persistence for debuggability - Thread-safe with RWMutex - Compatible with Temporal signal patterns - Non-destructive pause (snapshot before blocking) Closes T1.5
This commit is contained in:
Test
@@ -0,0 +1,261 @@
|
||||
package pause
|
||||
|
||||
import (
|
||||
"encoding/json"
|
||||
"fmt"
|
||||
"os"
|
||||
"path/filepath"
|
||||
"sync"
|
||||
"time"
|
||||
)
|
||||
|
||||
// WorkflowSnapshot represents a complete snapshot of workflow state
|
||||
type WorkflowSnapshot struct {
|
||||
WorkflowID string `json:"workflow_id"`
|
||||
Timestamp time.Time `json:"timestamp"`
|
||||
Stage string `json:"stage"`
|
||||
CompletedTasks []string `json:"completed_tasks"`
|
||||
PendingTasks []string `json:"pending_tasks"`
|
||||
FailedTasks []string `json:"failed_tasks"`
|
||||
CurrentTaskID string `json:"current_task_id"`
|
||||
CurrentActivityID string `json:"current_activity_id"`
|
||||
TaskMetrics map[string]interface{} `json:"task_metrics"`
|
||||
WorkflowMetrics map[string]interface{} `json:"workflow_metrics"`
|
||||
Configuration map[string]interface{} `json:"configuration"`
|
||||
Error string `json:"error,omitempty"`
|
||||
PausedAt time.Time `json:"paused_at"`
|
||||
ResumedAt *time.Time `json:"resumed_at,omitempty"`
|
||||
}
|
||||
|
||||
// SnapshotManager manages workflow state snapshots for pause/resume
|
||||
type SnapshotManager struct {
|
||||
mu sync.RWMutex
|
||||
basePath string
|
||||
snapshots map[string]*WorkflowSnapshot
|
||||
lastSnapshot *WorkflowSnapshot
|
||||
}
|
||||
|
||||
// NewSnapshotManager creates a new snapshot manager
|
||||
func NewSnapshotManager(basePath string) *SnapshotManager {
|
||||
return &SnapshotManager{
|
||||
basePath: basePath,
|
||||
snapshots: make(map[string]*WorkflowSnapshot),
|
||||
}
|
||||
}
|
||||
|
||||
// CreateSnapshot creates and persists a workflow snapshot
|
||||
func (sm *SnapshotManager) CreateSnapshot(
|
||||
workflowID string,
|
||||
stage string,
|
||||
completedTasks, pendingTasks, failedTasks []string,
|
||||
currentTaskID, currentActivityID string,
|
||||
taskMetrics, workflowMetrics, configuration map[string]interface{},
|
||||
) (*WorkflowSnapshot, error) {
|
||||
sm.mu.Lock()
|
||||
defer sm.mu.Unlock()
|
||||
|
||||
snapshot := &WorkflowSnapshot{
|
||||
WorkflowID: workflowID,
|
||||
Timestamp: time.Now(),
|
||||
Stage: stage,
|
||||
CompletedTasks: completedTasks,
|
||||
PendingTasks: pendingTasks,
|
||||
FailedTasks: failedTasks,
|
||||
CurrentTaskID: currentTaskID,
|
||||
CurrentActivityID: currentActivityID,
|
||||
TaskMetrics: taskMetrics,
|
||||
WorkflowMetrics: workflowMetrics,
|
||||
Configuration: configuration,
|
||||
PausedAt: time.Now(),
|
||||
}
|
||||
|
||||
sm.snapshots[workflowID] = snapshot
|
||||
sm.lastSnapshot = snapshot
|
||||
|
||||
return snapshot, sm.persistLocked(workflowID, snapshot)
|
||||
}
|
||||
|
||||
// GetLatestSnapshot retrieves the latest snapshot for a workflow
|
||||
func (sm *SnapshotManager) GetLatestSnapshot(workflowID string) *WorkflowSnapshot {
|
||||
sm.mu.RLock()
|
||||
defer sm.mu.RUnlock()
|
||||
|
||||
return sm.snapshots[workflowID]
|
||||
}
|
||||
|
||||
// HasSnapshot checks if a snapshot exists for a workflow
|
||||
func (sm *SnapshotManager) HasSnapshot(workflowID string) bool {
|
||||
sm.mu.RLock()
|
||||
defer sm.mu.RUnlock()
|
||||
|
||||
_, exists := sm.snapshots[workflowID]
|
||||
return exists
|
||||
}
|
||||
|
||||
// RestoreFromSnapshot restores workflow state from a snapshot
|
||||
func (sm *SnapshotManager) RestoreFromSnapshot(workflowID string) (*WorkflowSnapshot, error) {
|
||||
sm.mu.Lock()
|
||||
defer sm.mu.Unlock()
|
||||
|
||||
snapshot, exists := sm.snapshots[workflowID]
|
||||
if !exists {
|
||||
// Try to load from disk
|
||||
return nil, fmt.Errorf("no snapshot found for workflow: %s", workflowID)
|
||||
}
|
||||
|
||||
// Mark as resumed
|
||||
now := time.Now()
|
||||
snapshot.ResumedAt = &now
|
||||
|
||||
return snapshot, nil
|
||||
}
|
||||
|
||||
// MarkResumed updates a snapshot as resumed
|
||||
func (sm *SnapshotManager) MarkResumed(workflowID string) error {
|
||||
sm.mu.Lock()
|
||||
defer sm.mu.Unlock()
|
||||
|
||||
snapshot, exists := sm.snapshots[workflowID]
|
||||
if !exists {
|
||||
return fmt.Errorf("no snapshot found for workflow: %s", workflowID)
|
||||
}
|
||||
|
||||
now := time.Now()
|
||||
snapshot.ResumedAt = &now
|
||||
|
||||
return sm.persistLocked(workflowID, snapshot)
|
||||
}
|
||||
|
||||
// Load loads snapshots from disk
|
||||
func (sm *SnapshotManager) Load() error {
|
||||
sm.mu.Lock()
|
||||
defer sm.mu.Unlock()
|
||||
|
||||
snapshotDir := filepath.Join(sm.basePath, "snapshots")
|
||||
entries, err := os.ReadDir(snapshotDir)
|
||||
if err != nil {
|
||||
if os.IsNotExist(err) {
|
||||
return nil // Directory doesn't exist yet
|
||||
}
|
||||
return err
|
||||
}
|
||||
|
||||
for _, entry := range entries {
|
||||
if !entry.IsDir() && filepath.Ext(entry.Name()) == ".json" {
|
||||
data, err := os.ReadFile(filepath.Join(snapshotDir, entry.Name()))
|
||||
if err != nil {
|
||||
continue
|
||||
}
|
||||
|
||||
var snapshot WorkflowSnapshot
|
||||
if err := json.Unmarshal(data, &snapshot); err != nil {
|
||||
continue
|
||||
}
|
||||
|
||||
sm.snapshots[snapshot.WorkflowID] = &snapshot
|
||||
}
|
||||
}
|
||||
|
||||
return nil
|
||||
}
|
||||
|
||||
// persistLocked saves a snapshot to disk (must be called with lock held)
|
||||
func (sm *SnapshotManager) persistLocked(workflowID string, snapshot *WorkflowSnapshot) error {
|
||||
snapshotDir := filepath.Join(sm.basePath, "snapshots")
|
||||
|
||||
// Create directory if it doesn't exist
|
||||
if err := os.MkdirAll(snapshotDir, 0755); err != nil {
|
||||
return err
|
||||
}
|
||||
|
||||
snapshotPath := filepath.Join(snapshotDir, fmt.Sprintf("%s.snapshot.json", workflowID))
|
||||
|
||||
data, err := json.MarshalIndent(snapshot, "", " ")
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
|
||||
return os.WriteFile(snapshotPath, data, 0644)
|
||||
}
|
||||
|
||||
// DeleteSnapshot deletes a snapshot (after successful completion)
|
||||
func (sm *SnapshotManager) DeleteSnapshot(workflowID string) error {
|
||||
sm.mu.Lock()
|
||||
defer sm.mu.Unlock()
|
||||
|
||||
delete(sm.snapshots, workflowID)
|
||||
|
||||
snapshotPath := filepath.Join(sm.basePath, "snapshots", fmt.Sprintf("%s.snapshot.json", workflowID))
|
||||
if _, err := os.Stat(snapshotPath); err == nil {
|
||||
return os.Remove(snapshotPath)
|
||||
}
|
||||
|
||||
return nil
|
||||
}
|
||||
|
||||
// GetAllSnapshots returns all snapshots
|
||||
func (sm *SnapshotManager) GetAllSnapshots() []*WorkflowSnapshot {
|
||||
sm.mu.RLock()
|
||||
defer sm.mu.RUnlock()
|
||||
|
||||
snapshots := make([]*WorkflowSnapshot, 0, len(sm.snapshots))
|
||||
for _, snapshot := range sm.snapshots {
|
||||
snapshots = append(snapshots, snapshot)
|
||||
}
|
||||
|
||||
return snapshots
|
||||
}
|
||||
|
||||
// GetLastSnapshot returns the last snapshot created
|
||||
func (sm *SnapshotManager) GetLastSnapshot() *WorkflowSnapshot {
|
||||
sm.mu.RLock()
|
||||
defer sm.mu.RUnlock()
|
||||
|
||||
return sm.lastSnapshot
|
||||
}
|
||||
|
||||
// GetSnapshotStats returns statistics about snapshots
|
||||
func (sm *SnapshotManager) GetSnapshotStats() map[string]interface{} {
|
||||
sm.mu.RLock()
|
||||
defer sm.mu.RUnlock()
|
||||
|
||||
paused := 0
|
||||
resumed := 0
|
||||
|
||||
for _, snapshot := range sm.snapshots {
|
||||
if snapshot.ResumedAt != nil {
|
||||
resumed++
|
||||
} else {
|
||||
paused++
|
||||
}
|
||||
}
|
||||
|
||||
return map[string]interface{}{
|
||||
"total": len(sm.snapshots),
|
||||
"paused": paused,
|
||||
"resumed": resumed,
|
||||
}
|
||||
}
|
||||
|
||||
// ClearOldSnapshots removes snapshots older than the specified duration
|
||||
func (sm *SnapshotManager) ClearOldSnapshots(maxAge time.Duration) (int, error) {
|
||||
sm.mu.Lock()
|
||||
defer sm.mu.Unlock()
|
||||
|
||||
now := time.Now()
|
||||
toDelete := make([]string, 0)
|
||||
|
||||
for wfID, snapshot := range sm.snapshots {
|
||||
if now.Sub(snapshot.PausedAt) > maxAge {
|
||||
toDelete = append(toDelete, wfID)
|
||||
}
|
||||
}
|
||||
|
||||
for _, wfID := range toDelete {
|
||||
delete(sm.snapshots, wfID)
|
||||
snapshotPath := filepath.Join(sm.basePath, "snapshots", fmt.Sprintf("%s.snapshot.json", wfID))
|
||||
_ = os.Remove(snapshotPath)
|
||||
}
|
||||
|
||||
return len(toDelete), nil
|
||||
}
|
||||
Reference in New Issue
Block a user