Merge #157863

157863: roachtest/import: add concurrent import tests r=mw5h a=mw5h

Previously, the import-cancellation roachtest would load the entire TPC-H dataset in parallel. When this test was ported over to the new, fingerprint validating, import test we no longer needed to load the entire dataset to validate the import was successful (the old test simply ran the workload to ensure the data wasn't corrupt). For simplicity at the time, this seemed fine, but it did lose the test of concurrent imports into separate tables that the old test had. This patch adds concurrency back to the import/cancellation test and adds a separate import/concurrency test, both of which load three files in parallel.

Fixes: #157242
Release note: None

Co-authored-by: Matt White <[email protected]>

Author: craig[bot]

pkg/cmd/roachtest/tests/import.go

@@ -174,9 +174,20 @@ func allDatasets(_ *rand.Rand) []string {
 	return slices.Collect(maps.Keys(datasets))
 }
 
-func anyDataset(rng *rand.Rand) []string {
+func nDatasets(rng *rand.Rand, n int) []string {
 	allDatasets := allDatasets(rng)
-	return []string{allDatasets[rng.Intn(len(allDatasets))]}
+	rng.Shuffle(len(allDatasets), func(i, j int) {
+		allDatasets[i], allDatasets[j] = allDatasets[j], allDatasets[i]
+	})
+	return allDatasets[:n]
+}
+
+func anyThreeDatasets(rng *rand.Rand) []string {
+	return nDatasets(rng, 3)
+}
+
+func anyDataset(rng *rand.Rand) []string {
+	return nDatasets(rng, 1)
 }
 
 // importTestSpec represents a subtest within the import test.
@@ -232,6 +243,12 @@ var tests = []importTestSpec{
 		nodes:        []int{4},
 		datasetNames: One("tpch/lineitem"),
 	},
+	// Basic test importing three datasets concurrently.
+	{
+		subtestName:  "concurrency",
+		nodes:        []int{4},
+		datasetNames: FromFunc(anyThreeDatasets),
+	},
 	// Test with a decommissioned node.
 	{
 		subtestName:  "decommissioned",
@@ -280,7 +297,7 @@ var tests = []importTestSpec{
 	{
 		subtestName:  "cancellation",
 		nodes:        []int{4},
-		datasetNames: FromFunc(anyDataset),
+		datasetNames: FromFunc(anyThreeDatasets),
 		importRunner: importCancellationRunner,
 	},
 }