feat: implement loader error handling and refactor chunk processing in migration job

This commit is contained in:
2026-04-08 23:42:31 -05:00
parent 0ee5d9032c
commit dc632361e5
4 changed files with 134 additions and 40 deletions

View File

@@ -50,7 +50,7 @@ func processMigrationJob(sourceDb *sql.DB, targetDb *pgxpool.Pool, job Migration
extractorErrorHandler(ctx, chExtractorErrors, chBatches, chJobErrors)
}()
chChunks := make(chan Chunk, QueueSize)
chChunksRaw := make(chan Chunk, QueueSize)
maxExtractors := min(NumExtractors, len(batches))
var wgMssqlExtractors sync.WaitGroup
@@ -58,7 +58,7 @@ func processMigrationJob(sourceDb *sql.DB, targetDb *pgxpool.Pool, job Migration
extractStartTime := time.Now()
for range maxExtractors {
wgMssqlExtractors.Go(func() {
extractFromMssql(ctx, sourceDb, job, sourceColTypes, ChunkSize, chBatches, chChunks, chExtractorErrors, chJobErrors)
extractFromMssql(ctx, sourceDb, job, sourceColTypes, ChunkSize, chBatches, chChunksRaw, chExtractorErrors, chJobErrors)
})
}
@@ -72,41 +72,45 @@ func processMigrationJob(sourceDb *sql.DB, targetDb *pgxpool.Pool, job Migration
go func() {
wgMssqlExtractors.Wait()
close(chChunks)
close(chChunksRaw)
log.Infof("Extraction completed in %v", time.Since(extractStartTime))
}()
chChunksTransform := make(chan Chunk, QueueSize)
chChunksTransformed := make(chan Chunk, QueueSize)
var wgMssqlTransformers sync.WaitGroup
log.Infof("Starting %d MSSQL transformers...", maxExtractors)
transformStartTime := time.Now()
for range maxExtractors {
wgMssqlTransformers.Go(func() {
transformRowsMssql(ctx, sourceColTypes, chChunks, chChunksTransform, chJobErrors)
transformRowsMssql(ctx, sourceColTypes, chChunksRaw, chChunksTransformed, chJobErrors)
})
}
go func() {
wgMssqlTransformers.Wait()
close(chChunksTransform)
close(chChunksTransformed)
log.Infof("Transformation completed in %v", time.Since(transformStartTime))
}()
var wgPostgresLoaders sync.WaitGroup
chLoadersErrors := make(chan LoaderError)
go func() {
loaderErrorHandler(ctx, chLoadersErrors, chChunksTransformed, chJobErrors)
}()
log.Infof("Starting %d PostgreSQL loader(s)...", NumLoaders)
loaderStartTime := time.Now()
for range NumLoaders {
wgPostgresLoaders.Go(func() {
if err := loadRowsPostgres(ctx, job, targetColTypes, targetDb, chChunksTransform); err != nil {
log.Error("Unexpected error loading data into postgres: ", err)
}
loadRowsPostgres(ctx, targetDb, job, targetColTypes, chChunksTransformed, chLoadersErrors)
})
}
wgPostgresLoaders.Wait()
close(chLoadersErrors)
log.Infof("Loading completed in %v", time.Since(loaderStartTime))
totalDuration := time.Since(jobStartTime)