VDiff: add --no-samples option to vdiff show

changelog/25.0/25.0.0/summary.md

@@ -19,6 +19,7 @@
 - **[Minor Changes](#minor-changes)**
     - **[VReplication](#minor-changes-vreplication)**
         - [Default data protection for `_reverse` workflow cancel/complete](#vreplication-reverse-workflow-data-protection)
+        - [`vdiff show --no-samples` strips the per-table row-sample report](#vreplication-vdiff-no-samples)
         - [Preserve Materialize target data on cancel by default](#vreplication-materialize-cancel-data-protection)
     - **[VTGate](#minor-changes-vtgate)**
         - [Ingress bytes in query LogStats](#vtgate-logstats-ingress-bytes)
@@ -174,6 +175,14 @@ The `--keep-data` flag help text has been updated to note this default explicitl
 
 See [#19906](https://github.com/vitessio/vitess/pull/19906) for details.
 
+#### <a id="vreplication-vdiff-no-samples"/>`vdiff show --no-samples` strips the per-table row-sample report</a>
+
+`vtctldclient vdiff ... show` now accepts a `--no-samples` flag. When set, the per-table diff report has its row-sample arrays (`MismatchedRowsSample`, `ExtraRowsSourceSample`, `ExtraRowsTargetSample`) stripped on the tablet, while the scalar counters and all other summary fields are preserved. This avoids exceeding gRPC message limits when `vdiff show` aggregates large blob/JSON row samples across every target shard. It is exposed as `no_samples` on the `VDiffShowRequest` (vtctld) and `VDiffReportOptions` (tablet) protobuf messages, and is opt-in and backward compatible.
+
+`vdiff create --wait` also uses `no_samples` for its internal progress polls. Text output is unchanged; with `--format json`, the per-interval progress output no longer includes the row samples (they remain available via `vdiff show --verbose` once the diff completes).
+
+See [#20870](https://github.com/vitessio/vitess/pull/20870) for details.
+
 #### <a id="vreplication-materialize-cancel-data-protection"/>Preserve Materialize target data on cancel by default</a>
 
 `vtctldclient Materialize cancel` now preserves the materialized target tables and their data. To remove the target tables when canceling the workflow, explicitly pass `--keep-data=false`.

go/cmd/vtctldclient/command/vreplication/vdiff/vdiff.go

@@ -82,8 +82,9 @@ var (
 	}{}
 
 	showOptions = struct {
-		Arg     string
-		Verbose bool
+		Arg       string
+		Verbose   bool
+		NoSamples bool
 	}{}
 
 	stopOptions = struct {
@@ -322,6 +323,7 @@ func commandCreate(cmd *cobra.Command, args []string) error {
 					Workflow:       common.BaseOptions.Workflow,
 					TargetKeyspace: common.BaseOptions.TargetKeyspace,
 					Arg:            uuidStr,
+					NoSamples:      true,
 				})
 				if err != nil {
 					return err
@@ -645,6 +647,7 @@ func commandShow(cmd *cobra.Command, args []string) error {
 		Workflow:       common.BaseOptions.Workflow,
 		TargetKeyspace: common.BaseOptions.TargetKeyspace,
 		Arg:            showOptions.Arg,
+		NoSamples:      showOptions.NoSamples,
 	})
 	if err != nil {
 		return err
@@ -709,6 +712,7 @@ func registerCommands(root *cobra.Command) {
 	base.AddCommand(resume)
 
 	show.Flags().BoolVar(&showOptions.Verbose, "verbose", false, "Show verbose output in summaries")
+	show.Flags().BoolVar(&showOptions.NoSamples, "no-samples", false, "Strip the per-table diff report's row-sample arrays (keeping the scalar counters). Useful for large diffs where the samples can exceed gRPC message limits.")
 	base.AddCommand(show)
 
 	stop.Flags().StringSliceVar(&stopOptions.TargetShards, "target-shards", nil, "The target shards to stop the vdiff on; default is all shards.")

go/cmd/vtctldclient/command/vreplication/vdiff/vdiff_test.go

@@ -22,6 +22,7 @@ import (
 	"time"
 
 	"github.com/google/uuid"
+	"github.com/spf13/cobra"
 	"github.com/stretchr/testify/require"
 
 	"vitess.io/vitess/go/sqltypes"
@@ -813,3 +814,16 @@ func TestGetStructNames(t *testing.T) {
 	want := []string{"A", "B"}
 	require.Equal(t, want, got)
 }
+
+func TestShowNoSamplesFlag(t *testing.T) {
+	root := &cobra.Command{Use: "test"}
+	registerCommands(root)
+
+	showCmd, _, err := root.Find([]string{"VDiff", "show"})
+	require.NoError(t, err)
+
+	flag := showCmd.Flags().Lookup("no-samples")
+	require.NotNil(t, flag, "vdiff show must expose the --no-samples flag")
+	require.Equal(t, "false", flag.DefValue, "--no-samples must default to false")
+	require.Contains(t, flag.Usage, "gRPC message limits")
+}

go/vt/vtctl/workflow/vdiff.go

@@ -566,6 +566,11 @@ func (s *Server) VDiffShow(ctx context.Context, req *vtctldatapb.VDiffShowReques
 		Workflow:  req.Workflow,
 		Action:    string(vdiff.ShowAction),
 		ActionArg: req.Arg,
+		Options: &tabletmanagerdatapb.VDiffOptions{
+			ReportOptions: &tabletmanagerdatapb.VDiffReportOptions{
+				NoSamples: req.GetNoSamples(),
+			},
+		},
 	}
 
 	ts, err := s.buildTrafficSwitcher(ctx, req.TargetKeyspace, req.Workflow)

go/vt/vtctl/workflow/vdiff_test.go

@@ -754,3 +754,79 @@ func TestVDiffDelete(t *testing.T) {
 		})
 	}
 }
+
+func TestVDiffShow(t *testing.T) {
+	ctx := t.Context()
+	sourceKeyspace := &testKeyspace{
+		KeyspaceName: "sourceks",
+		ShardNames:   []string{"0"},
+	}
+	targetKeyspace := &testKeyspace{
+		KeyspaceName: "targetks",
+		ShardNames:   []string{"-80", "80-"},
+	}
+	workflow := "testwf"
+	uuid := uuid.New().String()
+	env := newTestEnv(t, ctx, defaultCellName, sourceKeyspace, targetKeyspace)
+	t.Cleanup(env.close)
+
+	env.tmc.strict = true
+	action := string(vdiff.ShowAction)
+
+	expectedShowRequest := func(noSamples bool) *tabletmanagerdatapb.VDiffRequest {
+		return &tabletmanagerdatapb.VDiffRequest{
+			Keyspace:  targetKeyspace.KeyspaceName,
+			Workflow:  workflow,
+			Action:    action,
+			ActionArg: uuid,
+			Options: &tabletmanagerdatapb.VDiffOptions{
+				ReportOptions: &tabletmanagerdatapb.VDiffReportOptions{
+					NoSamples: noSamples,
+				},
+			},
+		}
+	}
+	bothTargets := func(noSamples bool) map[*topodatapb.Tablet]*vdiffRequestResponse {
+		return map[*topodatapb.Tablet]*vdiffRequestResponse{
+			env.tablets[targetKeyspace.KeyspaceName][startingTargetTabletUID]:               {req: expectedShowRequest(noSamples)},
+			env.tablets[targetKeyspace.KeyspaceName][startingTargetTabletUID+tabletUIDStep]: {req: expectedShowRequest(noSamples)},
+		}
+	}
+
+	tests := []struct {
+		name                  string
+		req                   *vtctldatapb.VDiffShowRequest
+		expectedVDiffRequests map[*topodatapb.Tablet]*vdiffRequestResponse
+	}{
+		{
+			name: "default forwards no_samples=false",
+			req: &vtctldatapb.VDiffShowRequest{
+				TargetKeyspace: targetKeyspace.KeyspaceName,
+				Workflow:       workflow,
+				Arg:            uuid,
+			},
+			expectedVDiffRequests: bothTargets(false),
+		},
+		{
+			name: "no_samples forwards no_samples=true",
+			req: &vtctldatapb.VDiffShowRequest{
+				TargetKeyspace: targetKeyspace.KeyspaceName,
+				Workflow:       workflow,
+				Arg:            uuid,
+				NoSamples:      true,
+			},
+			expectedVDiffRequests: bothTargets(true),
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			for tab, vdr := range tt.expectedVDiffRequests {
+				env.tmc.expectVDiffRequest(tab, vdr)
+			}
+			got, err := env.ws.VDiffShow(ctx, tt.req)
+			require.NoError(t, err)
+			require.NotNil(t, got)
+			env.tmc.confirmVDiffRequests(t)
+		})
+	}
+}