Merge #60633

60633: release-20.2: kvserver: improve handling for removal of a replica, when multiple replicas already exist on the same node r=aayushshah15 a=lunevalex

Backport 1/1 commits from #60546.

/cc @cockroachdb/release

---

Fixes #60545

The allocator in some cases allows for a range to have a replica
on multiple stores of the same node. If that happens, it should allow
itself to fix the situation by removing one of the offending replicas.
This was only half working due to an ordering problem in how the replicas
appeared in the descriptor. It could remove the first replica, but not the second one.

.

Release note: None


Co-authored-by: Alex Lunev <alexl@cockroachlabs.com>

pkg/kv/kvserver/replica_command.go

@@ -1120,6 +1120,27 @@ func maybeLeaveAtomicChangeReplicasAndRemoveLearners(
 	return desc, nil
 }
 
+func validateReplicationChangesMultipleReplicasOnTheSameNode(
+	byStoreID map[roachpb.StoreID]roachpb.ReplicationChange,
+	rDescsForNode []roachpb.ReplicaDescriptor,
+) error {
+	if len(byStoreID) != 1 {
+		return errors.Errorf(
+			"An unexpected number of changes %s for range %s, the only valid operation when there are already multiple replicas on the same node is removal", byStoreID, rDescsForNode)
+	}
+	for _, rDesc := range rDescsForNode {
+		chg, ok := byStoreID[rDesc.StoreID]
+		if ok {
+			if chg.ChangeType != roachpb.REMOVE_REPLICA {
+				return errors.Errorf(
+					"Expected replica to be removed from %v instead got %v.", rDesc, chg)
+			}
+			return nil
+		}
+	}
+	return errors.Errorf("Expected a removal of one of the replicas in %s, instead got %s", rDescsForNode, byStoreID)
+}
+
 func validateReplicationChanges(
 	desc *roachpb.RangeDescriptor, chgs roachpb.ReplicationChanges,
 ) error {
@@ -1150,15 +1171,29 @@ func validateReplicationChanges(
 		byStoreID[chg.Target.StoreID] = chg
 	}
 
+	descriptorsByNodeID := make(map[roachpb.NodeID][]roachpb.ReplicaDescriptor, len(desc.Replicas().All()))
+	for _, rDesc := range desc.Replicas().All() {
+		descriptorsByNodeID[rDesc.NodeID] = append(descriptorsByNodeID[rDesc.NodeID], rDesc)
+	}
+
 	// Then, check that we're not adding a second replica on nodes that already
 	// have one, or "re-add" an existing replica. We delete from byNodeAndStoreID so that
 	// after this loop, it contains only Nodes that we haven't seen in desc.
-	for _, rDesc := range desc.Replicas().All() {
-		byStoreID, ok := byNodeAndStoreID[rDesc.NodeID]
+	for nodeID, rDescsForNode := range descriptorsByNodeID {
+		byStoreID, ok := byNodeAndStoreID[nodeID]
 		if !ok {
 			continue
 		}
-		delete(byNodeAndStoreID, rDesc.NodeID)
+		delete(byNodeAndStoreID, nodeID)
+		// The only valid thing to do when we already have multiple replicas on the
+		// same node is to remove a replica.
+		if len(rDescsForNode) > 1 {
+			if err := validateReplicationChangesMultipleReplicasOnTheSameNode(byStoreID, rDescsForNode); err != nil {
+				return err
+			}
+			continue
+		}
+		rDesc := rDescsForNode[0]
 		if len(byStoreID) == 2 {
 			chg, k := byStoreID[rDesc.StoreID]
 			// We should be removing the replica from the existing store during a

pkg/kv/kvserver/replica_command_test.go

@@ -198,28 +198,34 @@ func TestValidateReplicationChanges(t *testing.T) {
 		InternalReplicas: []roachpb.ReplicaDescriptor{
 			{NodeID: 1, StoreID: 1},
 			{NodeID: 2, StoreID: 2},
-			{NodeID: 1, StoreID: 2, Type: &learnerType},
+			{NodeID: 1, StoreID: 3, Type: &learnerType},
 		},
 	}
 	err = validateReplicationChanges(descRebalancing, roachpb.ReplicationChanges{
 		{ChangeType: roachpb.REMOVE_REPLICA, Target: roachpb.ReplicationTarget{NodeID: 1, StoreID: 1}},
 	})
 	require.NoError(t, err)
 
-	// Test Case 15: Do an add while rebalancing within a node
+	// Test Case 15: same as 14 but remove the second node
+	err = validateReplicationChanges(descRebalancing, roachpb.ReplicationChanges{
+		{ChangeType: roachpb.REMOVE_REPLICA, Target: roachpb.ReplicationTarget{NodeID: 1, StoreID: 3}},
+	})
+	require.NoError(t, err)
+
+	// Test Case 16: Do an add while rebalancing within a node
 	err = validateReplicationChanges(descRebalancing, roachpb.ReplicationChanges{
 		{ChangeType: roachpb.ADD_REPLICA, Target: roachpb.ReplicationTarget{NodeID: 3, StoreID: 3}},
 	})
 	require.NoError(t, err)
 
-	// Test Case 16: Remove/Add within a node is not allowed, since we expect Add/Remove
+	// Test Case 17: Remove/Add within a node is not allowed, since we expect Add/Remove
 	err = validateReplicationChanges(desc, roachpb.ReplicationChanges{
 		{ChangeType: roachpb.REMOVE_REPLICA, Target: roachpb.ReplicationTarget{NodeID: 1, StoreID: 1}},
 		{ChangeType: roachpb.ADD_REPLICA, Target: roachpb.ReplicationTarget{NodeID: 1, StoreID: 2}},
 	})
 	require.Regexp(t, "can only add-remove a replica within a node, but got ", err)
 
-	// Test Case 17: We are rebalancing within a node and have only one replica
+	// Test Case 18: We are rebalancing within a node and have only one replica
 	descSingle := &roachpb.RangeDescriptor{
 		InternalReplicas: []roachpb.ReplicaDescriptor{
 			{NodeID: 1, StoreID: 1},