go-migrate/cmd/go_migrate/process.go

package main

import (
	"context"
	"database/sql"
	"sync"
	"sync/atomic"
	"time"

	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/config"
	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/custom_errors"
	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/etl"
	"git.ksdemosapps.com/kylesoda/go-migrate/internal/app/models"
	"github.com/jackc/pgx/v5/pgxpool"
	_ "github.com/microsoft/go-mssqldb"
	log "github.com/sirupsen/logrus"
)

func processMigrationJob(
	ctx context.Context,
	sourceDb *sql.DB,
	targetDb *pgxpool.Pool,
	extractor etl.Extractor,
	transformer etl.Transformer,
	loader etl.Loader,
	job config.Job,
) JobResult {
	result := JobResult{
		JobName:   job.Name,
		StartTime: time.Now(),
	}

	var rowsRead, rowsLoaded, rowsFailed int64

	sourceColTypes, targetColTypes, err := GetColumnTypes(sourceDb, targetDb, job.SourceTable, job.TargetTable)
	if err != nil {
		result.Error = err
		return result
	}

	logColumnTypes(sourceColTypes, "Source col types")
	logColumnTypes(targetColTypes, "Target col types")

	jobCtx, cancel := context.WithCancel(ctx)
	defer cancel()

	partitions, err := partitionGeneratorMssql(jobCtx, sourceDb, job.SourceTable, job.RowsPerPartition)
	if err != nil {
		log.Error("Unexpected error calculating batch ranges: ", err)
	}

	chJobErrors := make(chan custom_errors.JobError, job.QueueSize)
	chExtractorErrors := make(chan custom_errors.ExtractorError, job.QueueSize)
	chLoadersErrors := make(chan custom_errors.LoaderError, job.QueueSize)
	chPartitions := make(chan models.Partition, job.QueueSize)
	chBatchesRaw := make(chan models.Batch, job.QueueSize)
	chBatchesTransformed := make(chan models.Batch, job.QueueSize)

	var wgActivePartitions sync.WaitGroup
	var wgActiveBatches sync.WaitGroup
	var wgExtractors sync.WaitGroup
	var wgTransformers sync.WaitGroup
	var wgLoaders sync.WaitGroup

	go func() {
		if err := custom_errors.JobErrorHandler(jobCtx, chJobErrors); err != nil {
			cancel()
			result.Error = err
		}
	}()

	go custom_errors.ExtractorErrorHandler(jobCtx, job.Retry.Attempts, chExtractorErrors, chPartitions, chJobErrors, &wgActivePartitions)
	go custom_errors.LoaderErrorHandler(jobCtx, job.Retry.Attempts, chLoadersErrors, chBatchesTransformed, chJobErrors, &wgActiveBatches)

	maxExtractors := min(job.MaxExtractors, len(partitions))
	log.Infof("Starting %d extractor(s)...", maxExtractors)

	for range maxExtractors {
		wgExtractors.Go(func() {
			extractor.Exec(
				jobCtx,
				job.SourceTable,
				sourceColTypes,
				job.BatchSize,
				chPartitions,
				chBatchesRaw,
				chExtractorErrors,
				chJobErrors,
				&wgActivePartitions,
				&rowsRead,
			)
		})
	}

	wgActivePartitions.Add(len(partitions))
	go func() {
		for _, batch := range partitions {
			chPartitions <- batch
		}
	}()

	log.Infof("Starting %d transformer(s)...", maxExtractors)

	for range maxExtractors {
		wgTransformers.Go(func() {
			transformer.Exec(
				jobCtx,
				sourceColTypes,
				chBatchesRaw,
				chBatchesTransformed,
				chJobErrors,
				&wgActiveBatches,
			)
		})
	}

	log.Infof("Starting %d loader(s)...", job.MaxLoaders)

	for range job.MaxLoaders {
		wgLoaders.Go(func() {
			loader.Exec(
				jobCtx,
				job.TargetTable,
				targetColTypes,
				chBatchesTransformed,
				chLoadersErrors,
				chJobErrors,
				&wgActiveBatches,
				&rowsLoaded,
			)
		})
	}

	go func() {
		wgActivePartitions.Wait()
		close(chPartitions)
		close(chExtractorErrors)

		wgExtractors.Wait()
		close(chBatchesRaw)

		wgTransformers.Wait()

		wgActiveBatches.Wait()
		close(chBatchesTransformed)
		close(chLoadersErrors)

		wgLoaders.Wait()

		cancel()
	}()

	<-jobCtx.Done()

	if ctx.Err() != nil {
		result.Error = ctx.Err()
	}

	result.Duration = time.Since(result.StartTime)
	result.RowsRead = atomic.LoadInt64(&rowsRead)
	result.RowsLoaded = atomic.LoadInt64(&rowsLoaded)
	result.RowsFailed = atomic.LoadInt64(&rowsFailed)

	return result
}

func logColumnTypes(columnTypes []models.ColumnType, label string) {
	log.Debug(label)

	for _, col := range columnTypes {
		log.Debugf("%+v", col)
	}
}