xlators default orders and volume default options

Xlator order for client volfile changed to,

```
debug/io-stats
performance/read-ahead
performance/io-threads
performance/nl-cache
performance/quick-read
performance/open-behind
performance/io-cache
performance/readdir-ahead
performance/write-behind
performance/md-cache
features/utime
features/shard
cluster/distribute
```

This PR also disables Self heal by default.(Fixes: #1485)

Signed-off-by: Aravinda VK <avishwan@redhat.com>

e2e/glustershd_test.go

@@ -208,6 +208,11 @@ func testGranularEntryHeal(t *testing.T, tc *testCluster) {
 	_, err := client.VolumeCreate(createReq)
 	r.Nil(err)
 
+	var optionReq api.VolOptionReq
+	optionReq.Options = map[string]string{"cluster/replicate.self-heal-daemon": "on"}
+	optionReq.AllowAdvanced = true
+	r.Nil(client.VolumeSet(volname, optionReq))
+
 	r.Nil(client.VolumeStart(volname, false), "volume start failed")
 
 	healInfo, err := client.SelfHealInfo(volname, "info-summary")
@@ -218,7 +223,6 @@ func testGranularEntryHeal(t *testing.T, tc *testCluster) {
 		}
 	}
 
-	var optionReq api.VolOptionReq
 	optionReq.Options = map[string]string{"cluster/replicate.granular-entry-heal": "enable"}
 	optionReq.AllowAdvanced = true
 	r.Nil(client.VolumeSet(volname, optionReq))
@@ -313,9 +317,13 @@ func testSplitBrainOperation(t *testing.T, tc *testCluster) {
 	_, err := client.VolumeCreate(createReq)
 	r.Nil(err)
 
+	var optionReq api.VolOptionReq
+	optionReq.Options = map[string]string{"cluster/replicate.self-heal-daemon": "on"}
+	optionReq.AllowAdvanced = true
+	r.Nil(client.VolumeSet(volname, optionReq))
+
 	r.Nil(client.VolumeStart(volname, false), "volume start failed")
 
-	var optionReq api.VolOptionReq
 	pidpath := path.Join(tc.gds[0].Rundir, "glustershd.pid")
 	optionReq.Options = map[string]string{"cluster/replicate.self-heal-daemon": "off"}
 	optionReq.AllowAdvanced = true

glusterd2/commands/volumes/grouped-options.go

@@ -9,14 +9,15 @@ var defaultGroupOptions = map[string]*api.OptionGroup{
 	"profile.default.replicate": {
 		Name: "profile.default.replicate",
 		Options: []api.VolumeOption{
-			{Name: "cluster/replicate.self-heal-daemon", OnValue: "on"},
-			{Name: "performance/md-cache", OnValue: "off"},
-			{Name: "performance/open-behind", OnValue: "off"},
+			{Name: "cluster/replicate.self-heal-daemon", OnValue: "off"},
 			{Name: "performance/quick-read", OnValue: "off"},
-			{Name: "performance/io-cache", OnValue: "off"},
+			{Name: "performance/io-cache", OnValue: "on"},
 			{Name: "performance/readdir-ahead", OnValue: "off"},
 			{Name: "performance/read-ahead", OnValue: "off"},
-			{Name: "performance/write-behind", OnValue: "off"},
+			{Name: "performance/write-behind", OnValue: "on"},
+			{Name: "performance/open-behind", OnValue: "on"},
+			{Name: "performance/md-cache", OnValue: "on"},
+			{Name: "performance/nl-cache", OnValue: "on"},
 		},
 		Description: "Default volume options for Replicate or Distributed Replicate volumes",
 	},
@@ -37,13 +38,14 @@ var defaultGroupOptions = map[string]*api.OptionGroup{
 	"profile.default.distribute": {
 		Name: "profile.default.distribute",
 		Options: []api.VolumeOption{
-			{Name: "performance/md-cache", OnValue: "off"},
-			{Name: "performance/open-behind", OnValue: "off"},
 			{Name: "performance/quick-read", OnValue: "off"},
-			{Name: "performance/io-cache", OnValue: "off"},
+			{Name: "performance/io-cache", OnValue: "on"},
 			{Name: "performance/readdir-ahead", OnValue: "off"},
 			{Name: "performance/read-ahead", OnValue: "off"},
-			{Name: "performance/write-behind", OnValue: "off"},
+			{Name: "performance/write-behind", OnValue: "on"},
+			{Name: "performance/open-behind", OnValue: "on"},
+			{Name: "performance/md-cache", OnValue: "on"},
+			{Name: "performance/nl-cache", OnValue: "on"},
 		},
 		Description: "Default volume options for Distribute volumes",
 	},

glusterd2/volgen/defaults.go

@@ -98,6 +98,25 @@ func init() {
 	}
 
 	// default client template
+
+	// Quick-read should be a parent of open-behind (otherwise
+	// we'll suffer perf penalty for small file reads)
+
+	// md-cache should be a descendant of write-behind (Otherwise
+	// we cannot leverage stats from brick in writev_cbk. Since
+	// writes invalidate kernel attributes, after a write, kernel
+	// will invariably ask glusterfs for stats. If stats are
+	// invalidated in md-cache by write-cbk - happens when
+	// md-cache is an ancestor of write-behind - stat fop will
+	// have to travel all the way to brick. However, if md-cache
+	// is a descendant of write-behind, stats in write-cbk from
+	// brick will be cached in md-cache)
+
+	// If client-io-threads is enabled, read-ahead should be
+	// parent of client-io-threads (parallelism introduced by
+	// client-io-threads messes the sequential read detection
+	// logic in read-ahead). So, better to change the relative
+	// order of client-io-threads and read-ahead too
 	tmpls[utils.ClientVolfile] = Template{
 		Name:  utils.ClientVolfile,
 		Level: VolfileLevelVolume,
@@ -107,28 +126,33 @@ func init() {
 				NameTmpl: "{{ volume.name }}",
 			},
 			{
-				Type: "performance/io-threads",
+				Type: "performance/read-ahead",
 			},
 			{
-				Type: "performance/md-cache",
+				Type:     "performance/io-threads",
+				Disabled: true,
 			},
 			{
-				Type: "performance/open-behind",
+				Type: "performance/nl-cache",
 			},
 			{
 				Type: "performance/quick-read",
 			},
+			{
+				Type: "performance/open-behind",
+			},
 			{
 				Type: "performance/io-cache",
 			},
 			{
 				Type: "performance/readdir-ahead",
 			},
+
 			{
-				Type: "performance/read-ahead",
+				Type: "performance/write-behind",
 			},
 			{
-				Type: "performance/write-behind",
+				Type: "performance/md-cache",
 			},
 			{
 				Type:           "features/read-only",