Merge #137067 #137094

137067: roachtest: update mt-upgrade test owner to db-server r=rimadeodhar a=rimadeodhar This PR updates the test ownership for the multitenant-upgrade test to the DB Server team. All future test failures will be routed to `t-db-server` for triage. Epic: none Release note: none 137094: roachtest: better triage schema workload flakes in backup-restore tests r=dt a=msbutler Informs #136709 Release note: none Co-authored-by: rimadeodhar <[email protected]> Co-authored-by: Michael Butler <[email protected]>
cockroachdb · Dec 10, 2024 · 46e8e0f · 46e8e0f
3 parents 95d95a6 + 139bbcd + b0632b9
commit 46e8e0f
Show file tree

Hide file tree

Showing 3 changed files with 15 additions and 5 deletions.
diff --git a/pkg/cmd/roachtest/registry/owners.go b/pkg/cmd/roachtest/registry/owners.go
@@ -25,7 +25,7 @@ const (
 	OwnerAdmissionControl   Owner = `admission-control`
 	OwnerObservability      Owner = `obs-prs`
 	OwnerObservabilityIndia Owner = `obs-india-prs`
-	OwnerServer             Owner = `server` // not currently staffed
+	OwnerServer             Owner = `server`
 	OwnerSQLFoundations     Owner = `sql-foundations`
 	OwnerMigrations         Owner = `migrations`
 	OwnerProductSecurity    Owner = `product-security`

diff --git a/pkg/cmd/roachtest/tests/backup_restore_roundtrip.go b/pkg/cmd/roachtest/tests/backup_restore_roundtrip.go
@@ -10,6 +10,7 @@ import (
 	gosql "database/sql"
 	"fmt"
 	"math/rand"
+	"strings"
 	"time"
 
 	"github.com/cockroachdb/cockroach/pkg/cmd/roachtest/cluster"
@@ -254,17 +255,26 @@ func startBackgroundWorkloads(
 	if err != nil {
 		return nil, err
 	}
+
+	handleChemaChangeError := func(err error) error {
+		// If the UNEXPECTED ERROR detail appears, the workload likely flaked.
+		// Otherwise, the workload could have failed due to other reasons like a node
+		// crash.
+		if err != nil && strings.Contains(errors.FlattenDetails(err), "UNEXPECTED ERROR") {
+			return registry.ErrorWithOwner(registry.OwnerSQLFoundations, errors.Wrapf(err, "schema change workload failed"))
+		}
+		return err
+	}
 	err = c.RunE(ctx, option.WithNodes(workloadNode), scInit.String())
 	if err != nil {
-		return nil, registry.ErrorWithOwner(registry.OwnerSQLFoundations, errors.Wrapf(err, "failed to init schema change workload"))
+		return nil, handleChemaChangeError(err)
 	}
 
 	run := func() (func(), error) {
 		tables, err := testUtils.loadTablesForDBs(ctx, l, testRNG, dbs...)
 		if err != nil {
 			return nil, err
 		}
-
 		stopBank := workloadWithCancel(m, func(ctx context.Context) error {
 			return c.RunE(ctx, option.WithNodes(workloadNode), bankRun.String())
 		})
@@ -274,7 +284,7 @@ func startBackgroundWorkloads(
 		})
 		stopSC := workloadWithCancel(m, func(ctx context.Context) error {
 			if err := c.RunE(ctx, option.WithNodes(workloadNode), scRun.String()); err != nil {
-				return registry.ErrorWithOwner(registry.OwnerSQLFoundations, errors.Wrapf(err, "failed to run schema change workload"))
+				return handleChemaChangeError(err)
 			}
 			return nil
 		})

diff --git a/pkg/cmd/roachtest/tests/multitenant_upgrade.go b/pkg/cmd/roachtest/tests/multitenant_upgrade.go
@@ -32,7 +32,7 @@ func registerMultiTenantUpgrade(r registry.Registry) {
 		Cluster:          r.MakeClusterSpec(7),
 		CompatibleClouds: registry.CloudsWithServiceRegistration,
 		Suites:           registry.Suites(registry.Nightly),
-		Owner:            registry.OwnerDisasterRecovery,
+		Owner:            registry.OwnerServer,
 		Run: func(ctx context.Context, t test.Test, c cluster.Cluster) {
 			runMultitenantUpgrade(ctx, t, c)
 		},