Files
go-migrate/cmd/go_migrate/loader.go

172 lines
3.7 KiB
Go

package main
import (
"context"
"database/sql"
"fmt"
"sync"
"time"
"github.com/jackc/pgx/v5"
"github.com/jackc/pgx/v5/pgxpool"
mssql "github.com/microsoft/go-mssqldb"
log "github.com/sirupsen/logrus"
)
func loadRowsPostgres(
ctx context.Context,
db *pgxpool.Pool,
job MigrationJob,
columns []ColumnType,
chChunksIn <-chan Chunk,
chErrorsOut chan<- LoaderError,
wgActiveChunks *sync.WaitGroup,
) {
tableId := pgx.Identifier{job.Schema, job.Table}
colNames := Map(columns, func(col ColumnType) string {
return col.name
})
for {
if ctx.Err() != nil {
return
}
select {
case <-ctx.Done():
return
case chunk, ok := <-chChunksIn:
if !ok {
return
}
if abort := loadChunkPostgres(ctx, db, tableId, colNames, chunk, chErrorsOut, wgActiveChunks); abort {
return
}
}
}
}
func loadChunkPostgres(
ctx context.Context,
db *pgxpool.Pool,
identifier pgx.Identifier,
colNames []string,
chunk Chunk,
chErrorsOut chan<- LoaderError,
wgActiveChunks *sync.WaitGroup,
) (abort bool) {
chunkStartTime := time.Now()
_, err := db.CopyFrom(
ctx,
identifier,
colNames,
pgx.CopyFromRows(chunk.Data),
)
if err != nil {
select {
case chErrorsOut <- LoaderError{Chunk: chunk, Msg: err.Error()}:
case <-ctx.Done():
return true
}
return false
}
chunkDuration := time.Since(chunkStartTime)
rowsPerSec := float64(len(chunk.Data)) / chunkDuration.Seconds()
log.Infof("Loaded chunk: %d rows in %v (%.0f rows/sec)", len(chunk.Data), chunkDuration, rowsPerSec)
wgActiveChunks.Done()
return false
}
func loadRowsMssql(ctx context.Context, job MigrationJob, columns []ColumnType, db *sql.DB, in <-chan []UnknownRowValues) error {
chunkCount := 0
totalRowsLoaded := 0
for rows := range in {
chunkStartTime := time.Now()
tx, err := db.BeginTx(ctx, nil)
if err != nil {
return fmt.Errorf("error starting transaction: %w", err)
}
fullTableName := fmt.Sprintf("[%s].[%s]", job.Schema, job.Table)
colNames := Map(columns, func(col ColumnType) string {
return col.name
})
stmt, err := tx.PrepareContext(ctx, mssql.CopyIn(fullTableName, mssql.BulkOptions{}, colNames...))
if err != nil {
tx.Rollback()
return fmt.Errorf("error preparing bulk copy statement: %w", err)
}
copyStartTime := time.Now()
for _, row := range rows {
_, err = stmt.ExecContext(ctx, row...)
if err != nil {
stmt.Close()
tx.Rollback()
return fmt.Errorf("error executing row insert: %w", err)
}
}
result, err := stmt.ExecContext(ctx)
if err != nil {
stmt.Close()
tx.Rollback()
return fmt.Errorf("error flushing bulk data: %w", err)
}
err = stmt.Close()
if err != nil {
tx.Rollback()
return fmt.Errorf("error closing statement: %w", err)
}
if err := tx.Commit(); err != nil {
return fmt.Errorf("error committing transaction: %w", err)
}
rowsAffected, _ := result.RowsAffected()
chunkCount++
totalRowsLoaded += int(rowsAffected)
copyDuration := time.Since(copyStartTime)
chunkDuration := time.Since(chunkStartTime)
rowsPerSec := float64(len(rows)) / chunkDuration.Seconds()
log.Infof("Loaded chunk #%d (MSSQL): %d rows in %v (copy: %v, %.0f rows/sec) - Total: %d rows", chunkCount, len(rows), chunkDuration, copyDuration, rowsPerSec, totalRowsLoaded)
}
return nil
}
func Map[T any, V any](input []T, mapper func(T) V) []V {
result := make([]V, len(input))
for i, v := range input {
result[i] = mapper(v)
}
return result
}
func fakeLoader(job MigrationJob, columns []ColumnType, in <-chan [][]any) {
for rows := range in {
log.Debugf("Chunk received, loading data into...")
for i, rowValues := range rows {
if i%100 == 0 {
logSampleRow(job, columns, rowValues, fmt.Sprintf("row %d", i))
}
}
}
}