Add tools for archive manager and archive-migration

Needs testing and review

tools/archive-manager/import_test.go

@@ -0,0 +1,335 @@
+// Copyright (C) NHR@FAU, University Erlangen-Nuremberg.
+// All rights reserved. This file is part of cc-backend.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+package main
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/ClusterCockpit/cc-backend/pkg/archive"
+	"github.com/ClusterCockpit/cc-lib/schema"
+	"github.com/ClusterCockpit/cc-lib/util"
+)
+
+// TestImportFileToSqlite tests importing jobs from file backend to SQLite backend
+func TestImportFileToSqlite(t *testing.T) {
+	// Create temporary directories
+	tmpdir := t.TempDir()
+	srcArchive := filepath.Join(tmpdir, "src-archive")
+	dstDb := filepath.Join(tmpdir, "dst-archive.db")
+
+	// Copy test data to source archive
+	testDataPath := "../../pkg/archive/testdata/archive"
+	if _, err := os.Stat(testDataPath); os.IsNotExist(err) {
+		t.Skip("Test data not found, skipping integration test")
+	}
+
+	if err := util.CopyDir(testDataPath, srcArchive); err != nil {
+		t.Fatalf("Failed to copy test data: %s", err.Error())
+	}
+
+	// Initialize source backend (file)
+	srcConfig := fmt.Sprintf(`{"kind":"file","path":"%s"}`, srcArchive)
+	srcBackend, err := archive.InitBackend(json.RawMessage(srcConfig))
+	if err != nil {
+		t.Fatalf("Failed to initialize source backend: %s", err.Error())
+	}
+
+	// Initialize destination backend (sqlite)
+	dstConfig := fmt.Sprintf(`{"kind":"sqlite","dbPath":"%s"}`, dstDb)
+	dstBackend, err := archive.InitBackend(json.RawMessage(dstConfig))
+	if err != nil {
+		t.Fatalf("Failed to initialize destination backend: %s", err.Error())
+	}
+
+	// Perform import
+	imported, failed, err := importArchive(srcBackend, dstBackend)
+	if err != nil {
+		t.Errorf("Import failed: %s", err.Error())
+	}
+
+	if imported == 0 {
+		t.Error("No jobs were imported")
+	}
+
+	if failed > 0 {
+		t.Errorf("%d jobs failed to import", failed)
+	}
+
+	t.Logf("Successfully imported %d jobs", imported)
+
+	// Verify jobs exist in destination
+	// Count jobs in source
+	srcCount := 0
+	for range srcBackend.Iter(false) {
+		srcCount++
+	}
+
+	// Count jobs in destination
+	dstCount := 0
+	for range dstBackend.Iter(false) {
+		dstCount++
+	}
+
+	if srcCount != dstCount {
+		t.Errorf("Job count mismatch: source has %d jobs, destination has %d jobs", srcCount, dstCount)
+	}
+}
+
+// TestImportFileToFile tests importing jobs from one file backend to another
+func TestImportFileToFile(t *testing.T) {
+	// Create temporary directories
+	tmpdir := t.TempDir()
+	srcArchive := filepath.Join(tmpdir, "src-archive")
+	dstArchive := filepath.Join(tmpdir, "dst-archive")
+
+	// Copy test data to source archive
+	testDataPath := "../../pkg/archive/testdata/archive"
+	if _, err := os.Stat(testDataPath); os.IsNotExist(err) {
+		t.Skip("Test data not found, skipping integration test")
+	}
+
+	if err := util.CopyDir(testDataPath, srcArchive); err != nil {
+		t.Fatalf("Failed to copy test data: %s", err.Error())
+	}
+
+	// Create destination archive directory
+	if err := os.MkdirAll(dstArchive, 0755); err != nil {
+		t.Fatalf("Failed to create destination directory: %s", err.Error())
+	}
+
+	// Initialize source backend
+	srcConfig := fmt.Sprintf(`{"kind":"file","path":"%s"}`, srcArchive)
+	srcBackend, err := archive.InitBackend(json.RawMessage(srcConfig))
+	if err != nil {
+		t.Fatalf("Failed to initialize source backend: %s", err.Error())
+	}
+
+	// Initialize destination backend
+	dstConfig := fmt.Sprintf(`{"kind":"file","path":"%s"}`, dstArchive)
+	dstBackend, err := archive.InitBackend(json.RawMessage(dstConfig))
+	if err != nil {
+		t.Fatalf("Failed to initialize destination backend: %s", err.Error())
+	}
+
+	// Perform import
+	imported, failed, err := importArchive(srcBackend, dstBackend)
+	if err != nil {
+		t.Errorf("Import failed: %s", err.Error())
+	}
+
+	if imported == 0 {
+		t.Error("No jobs were imported")
+	}
+
+	if failed > 0 {
+		t.Errorf("%d jobs failed to import", failed)
+	}
+
+	t.Logf("Successfully imported %d jobs", imported)
+}
+
+// TestImportDataIntegrity verifies that job metadata and data are correctly imported
+func TestImportDataIntegrity(t *testing.T) {
+	// Create temporary directories
+	tmpdir := t.TempDir()
+	srcArchive := filepath.Join(tmpdir, "src-archive")
+	dstDb := filepath.Join(tmpdir, "dst-archive.db")
+
+	// Copy test data to source archive
+	testDataPath := "../../pkg/archive/testdata/archive"
+	if _, err := os.Stat(testDataPath); os.IsNotExist(err) {
+		t.Skip("Test data not found, skipping integration test")
+	}
+
+	if err := util.CopyDir(testDataPath, srcArchive); err != nil {
+		t.Fatalf("Failed to copy test data: %s", err.Error())
+	}
+
+	// Initialize backends
+	srcConfig := fmt.Sprintf(`{"kind":"file","path":"%s"}`, srcArchive)
+	srcBackend, err := archive.InitBackend(json.RawMessage(srcConfig))
+	if err != nil {
+		t.Fatalf("Failed to initialize source backend: %s", err.Error())
+	}
+
+	dstConfig := fmt.Sprintf(`{"kind":"sqlite","dbPath":"%s"}`, dstDb)
+	dstBackend, err := archive.InitBackend(json.RawMessage(dstConfig))
+	if err != nil {
+		t.Fatalf("Failed to initialize destination backend: %s", err.Error())
+	}
+
+	// Perform import
+	_, _, err = importArchive(srcBackend, dstBackend)
+	if err != nil {
+		t.Errorf("Import failed: %s", err.Error())
+	}
+
+	// Verify data integrity for each job
+	verifiedJobs := 0
+	for srcJob := range srcBackend.Iter(false) {
+		if srcJob.Meta == nil {
+			continue
+		}
+
+		// Load job from destination
+		dstJobMeta, err := dstBackend.LoadJobMeta(srcJob.Meta)
+		if err != nil {
+			t.Errorf("Failed to load job %d from destination: %s", srcJob.Meta.JobID, err.Error())
+			continue
+		}
+
+		// Verify basic metadata
+		if dstJobMeta.JobID != srcJob.Meta.JobID {
+			t.Errorf("JobID mismatch: expected %d, got %d", srcJob.Meta.JobID, dstJobMeta.JobID)
+		}
+
+		if dstJobMeta.Cluster != srcJob.Meta.Cluster {
+			t.Errorf("Cluster mismatch for job %d: expected %s, got %s",
+				srcJob.Meta.JobID, srcJob.Meta.Cluster, dstJobMeta.Cluster)
+		}
+
+		if dstJobMeta.StartTime != srcJob.Meta.StartTime {
+			t.Errorf("StartTime mismatch for job %d: expected %d, got %d",
+				srcJob.Meta.JobID, srcJob.Meta.StartTime, dstJobMeta.StartTime)
+		}
+
+		// Load and verify job data
+		srcData, err := srcBackend.LoadJobData(srcJob.Meta)
+		if err != nil {
+			t.Errorf("Failed to load job data from source: %s", err.Error())
+			continue
+		}
+
+		dstData, err := dstBackend.LoadJobData(srcJob.Meta)
+		if err != nil {
+			t.Errorf("Failed to load job data from destination: %s", err.Error())
+			continue
+		}
+
+		// Verify metric data exists
+		if len(srcData) != len(dstData) {
+			t.Errorf("Metric count mismatch for job %d: expected %d, got %d",
+				srcJob.Meta.JobID, len(srcData), len(dstData))
+		}
+
+		verifiedJobs++
+	}
+
+	if verifiedJobs == 0 {
+		t.Error("No jobs were verified")
+	}
+
+	t.Logf("Successfully verified %d jobs", verifiedJobs)
+}
+
+// TestImportEmptyArchive tests importing from an empty archive
+func TestImportEmptyArchive(t *testing.T) {
+	tmpdir := t.TempDir()
+	srcArchive := filepath.Join(tmpdir, "empty-archive")
+	dstDb := filepath.Join(tmpdir, "dst-archive.db")
+
+	// Create empty source archive
+	if err := os.MkdirAll(srcArchive, 0755); err != nil {
+		t.Fatalf("Failed to create source directory: %s", err.Error())
+	}
+
+	// Write version file
+	versionFile := filepath.Join(srcArchive, "version.txt")
+	if err := os.WriteFile(versionFile, []byte("3"), 0644); err != nil {
+		t.Fatalf("Failed to write version file: %s", err.Error())
+	}
+
+	// Initialize backends
+	srcConfig := fmt.Sprintf(`{"kind":"file","path":"%s"}`, srcArchive)
+	srcBackend, err := archive.InitBackend(json.RawMessage(srcConfig))
+	if err != nil {
+		t.Fatalf("Failed to initialize source backend: %s", err.Error())
+	}
+
+	dstConfig := fmt.Sprintf(`{"kind":"sqlite","dbPath":"%s"}`, dstDb)
+	dstBackend, err := archive.InitBackend(json.RawMessage(dstConfig))
+	if err != nil {
+		t.Fatalf("Failed to initialize destination backend: %s", err.Error())
+	}
+
+	// Perform import
+	imported, failed, err := importArchive(srcBackend, dstBackend)
+	if err != nil {
+		t.Errorf("Import from empty archive should not fail: %s", err.Error())
+	}
+
+	if imported != 0 {
+		t.Errorf("Expected 0 imported jobs, got %d", imported)
+	}
+
+	if failed != 0 {
+		t.Errorf("Expected 0 failed jobs, got %d", failed)
+	}
+}
+
+// TestImportDuplicateJobs tests that duplicate jobs are skipped
+func TestImportDuplicateJobs(t *testing.T) {
+	tmpdir := t.TempDir()
+	srcArchive := filepath.Join(tmpdir, "src-archive")
+	dstDb := filepath.Join(tmpdir, "dst-archive.db")
+
+	// Copy test data
+	testDataPath := "../../pkg/archive/testdata/archive"
+	if _, err := os.Stat(testDataPath); os.IsNotExist(err) {
+		t.Skip("Test data not found, skipping integration test")
+	}
+
+	if err := util.CopyDir(testDataPath, srcArchive); err != nil {
+		t.Fatalf("Failed to copy test data: %s", err.Error())
+	}
+
+	// Initialize backends
+	srcConfig := fmt.Sprintf(`{"kind":"file","path":"%s"}`, srcArchive)
+	srcBackend, err := archive.InitBackend(json.RawMessage(srcConfig))
+	if err != nil {
+		t.Fatalf("Failed to initialize source backend: %s", err.Error())
+	}
+
+	dstConfig := fmt.Sprintf(`{"kind":"sqlite","dbPath":"%s"}`, dstDb)
+	dstBackend, err := archive.InitBackend(json.RawMessage(dstConfig))
+	if err != nil {
+		t.Fatalf("Failed to initialize destination backend: %s", err.Error())
+	}
+
+	// First import
+	imported1, _, err := importArchive(srcBackend, dstBackend)
+	if err != nil {
+		t.Fatalf("First import failed: %s", err.Error())
+	}
+
+	// Second import (should skip all jobs)
+	imported2, _, err := importArchive(srcBackend, dstBackend)
+	if err != nil {
+		t.Errorf("Second import failed: %s", err.Error())
+	}
+
+	if imported2 != 0 {
+		t.Errorf("Second import should skip all jobs, but imported %d", imported2)
+	}
+
+	t.Logf("First import: %d jobs, Second import: %d jobs (all skipped as expected)", imported1, imported2)
+}
+
+// TestJobStub is a helper test to verify that the job stub used in tests matches the schema
+func TestJobStub(t *testing.T) {
+	job := &schema.Job{
+		JobID:     123,
+		Cluster:   "test-cluster",
+		StartTime: 1234567890,
+	}
+
+	if job.JobID != 123 {
+		t.Errorf("Expected JobID 123, got %d", job.JobID)
+	}
+}

tools/archive-manager/main.go

@@ -9,6 +9,8 @@ import (
 	"flag"
 	"fmt"
 	"os"
+	"sync"
+	"sync/atomic"
 	"time"
 
 	"github.com/ClusterCockpit/cc-backend/internal/config"
@@ -31,9 +33,106 @@ func parseDate(in string) int64 {
 	return 0
 }
 
+// importArchive imports all jobs from a source archive backend to a destination archive backend.
+// It uses parallel processing with a worker pool to improve performance.
+// Returns the number of successfully imported jobs, failed jobs, and any error encountered.
+func importArchive(srcBackend, dstBackend archive.ArchiveBackend) (int, int, error) {
+	cclog.Info("Starting parallel archive import...")
+
+	// Use atomic counters for thread-safe updates
+	var imported int32
+	var failed int32
+	var skipped int32
+
+	// Number of parallel workers
+	numWorkers := 4
+	cclog.Infof("Using %d parallel workers", numWorkers)
+
+	// Create channels for job distribution
+	jobs := make(chan archive.JobContainer, numWorkers*2)
+	
+	// WaitGroup to track worker completion
+	var wg sync.WaitGroup
+
+	// Start worker goroutines
+	for i := 0; i < numWorkers; i++ {
+		wg.Add(1)
+		go func(workerID int) {
+			defer wg.Done()
+
+			for job := range jobs {
+				// Validate job metadata
+				if job.Meta == nil {
+					cclog.Warn("Skipping job with nil metadata")
+					atomic.AddInt32(&failed, 1)
+					continue
+				}
+
+				// Validate job data
+				if job.Data == nil {
+					cclog.Warnf("Job %d from cluster %s has no metric data, skipping",
+						job.Meta.JobID, job.Meta.Cluster)
+					atomic.AddInt32(&failed, 1)
+					continue
+				}
+
+				// Check if job already exists in destination
+				if dstBackend.Exists(job.Meta) {
+					cclog.Debugf("Job %d (cluster: %s, start: %d) already exists in destination, skipping",
+						job.Meta.JobID, job.Meta.Cluster, job.Meta.StartTime)
+					atomic.AddInt32(&skipped, 1)
+					continue
+				}
+
+				// Import job to destination
+				if err := dstBackend.ImportJob(job.Meta, job.Data); err != nil {
+					cclog.Errorf("Failed to import job %d from cluster %s: %s",
+						job.Meta.JobID, job.Meta.Cluster, err.Error())
+					atomic.AddInt32(&failed, 1)
+					continue
+				}
+
+				// Successfully imported
+				newCount := atomic.AddInt32(&imported, 1)
+				if newCount%100 == 0 {
+					cclog.Infof("Progress: %d jobs imported, %d skipped, %d failed",
+						newCount, atomic.LoadInt32(&skipped), atomic.LoadInt32(&failed))
+				}
+			}
+		}(i)
+	}
+
+	// Feed jobs to workers
+	go func() {
+		for job := range srcBackend.Iter(true) {
+			jobs <- job
+		}
+		close(jobs)
+	}()
+
+	// Wait for all workers to complete
+	wg.Wait()
+
+	finalImported := int(atomic.LoadInt32(&imported))
+	finalFailed := int(atomic.LoadInt32(&failed))
+	finalSkipped := int(atomic.LoadInt32(&skipped))
+
+	cclog.Infof("Import completed: %d jobs imported, %d skipped, %d failed",
+		finalImported, finalSkipped, finalFailed)
+
+	if finalFailed > 0 {
+		return finalImported, finalFailed, fmt.Errorf("%d jobs failed to import", finalFailed)
+	}
+
+	return finalImported, finalFailed, nil
+}
+
+
+
 func main() {
 	var srcPath, flagConfigFile, flagLogLevel, flagRemoveCluster, flagRemoveAfter, flagRemoveBefore string
-	var flagLogDateTime, flagValidate bool
+	var flagSrcConfig, flagDstConfig string
+	var flagLogDateTime, flagValidate, flagImport bool
 
 	flag.StringVar(&srcPath, "s", "./var/job-archive", "Specify the source job archive path. Default is ./var/job-archive")
 	flag.BoolVar(&flagLogDateTime, "logdate", false, "Set this flag to add date and time to log messages")
@@ -43,14 +142,54 @@ func main() {
 	flag.StringVar(&flagRemoveBefore, "remove-before", "", "Remove all jobs with start time before date (Format: 2006-Jan-04)")
 	flag.StringVar(&flagRemoveAfter, "remove-after", "", "Remove all jobs with start time after date (Format: 2006-Jan-04)")
 	flag.BoolVar(&flagValidate, "validate", false, "Set this flag to validate a job archive against the json schema")
+	flag.BoolVar(&flagImport, "import", false, "Import jobs from source archive to destination archive")
+	flag.StringVar(&flagSrcConfig, "src-config", "", "Source archive backend configuration (JSON), e.g. '{\"kind\":\"file\",\"path\":\"./archive\"}'")
+	flag.StringVar(&flagDstConfig, "dst-config", "", "Destination archive backend configuration (JSON), e.g. '{\"kind\":\"sqlite\",\"dbPath\":\"./archive.db\"}'")
 	flag.Parse()
 
+
 	archiveCfg := fmt.Sprintf("{\"kind\": \"file\",\"path\": \"%s\"}", srcPath)
 
 	cclog.Init(flagLogLevel, flagLogDateTime)
 
+	// Handle import mode
+	if flagImport {
+		if flagSrcConfig == "" || flagDstConfig == "" {
+			cclog.Fatal("Both --src-config and --dst-config must be specified for import mode")
+		}
+
+		cclog.Info("Import mode: initializing source and destination backends...")
+
+		// Initialize source backend
+		srcBackend, err := archive.InitBackend(json.RawMessage(flagSrcConfig))
+		if err != nil {
+			cclog.Fatalf("Failed to initialize source backend: %s", err.Error())
+		}
+		cclog.Info("Source backend initialized successfully")
+
+		// Initialize destination backend
+		dstBackend, err := archive.InitBackend(json.RawMessage(flagDstConfig))
+		if err != nil {
+			cclog.Fatalf("Failed to initialize destination backend: %s", err.Error())
+		}
+		cclog.Info("Destination backend initialized successfully")
+
+		// Perform import
+		imported, failed, err := importArchive(srcBackend, dstBackend)
+		if err != nil {
+			cclog.Errorf("Import completed with errors: %s", err.Error())
+			if failed > 0 {
+				os.Exit(1)
+			}
+		}
+
+		cclog.Infof("Import finished successfully: %d jobs imported", imported)
+		os.Exit(0)
+	}
+
 	ccconf.Init(flagConfigFile)
 
+
 	// Load and check main configuration
 	if cfg := ccconf.GetPackageConfig("main"); cfg != nil {
 		if clustercfg := ccconf.GetPackageConfig("clusters"); clustercfg != nil {