Implement MySQL version upgrade support

pkg/controllers/cluster/controller.go

@@ -17,6 +17,7 @@ package cluster
 import (
 	"context"
 	"fmt"
+	"strings"
 	"time"
 
 	apps "k8s.io/api/apps/v1beta1"
@@ -38,6 +39,7 @@ import (
 	record "k8s.io/client-go/tools/record"
 	workqueue "k8s.io/client-go/util/workqueue"
 
+	"github.com/coreos/go-semver/semver"
 	"github.com/golang/glog"
 	"github.com/pkg/errors"
 
@@ -390,9 +392,13 @@ func (m *MySQLController) syncHandler(key string) error {
 	}
 
 	// Upgrade the required component resources the current MySQLOperator version.
-	err = m.ensureMySQLOperatorVersion(cluster, ss, buildversion.GetBuildVersion())
-	if err != nil {
-		return err
+	if err := m.ensureMySQLOperatorVersion(cluster, ss, buildversion.GetBuildVersion()); err != nil {
+		return errors.Wrap(err, "ensuring MySQL Operator version")
+	}
+
+	// Upgrade the MySQL server version if required.
+	if err := m.ensureMySQLVersion(cluster, ss); err != nil {
+		return errors.Wrap(err, "ensuring MySQL version")
 	}
 
 	// If this number of the members on the Cluster does not equal the
@@ -421,6 +427,67 @@ func (m *MySQLController) syncHandler(key string) error {
 	return nil
 }
 
+func getMySQLContainerIndex(containers []corev1.Container) (int, error) {
+	for i, c := range containers {
+		if c.Name == statefulsets.MySQLServerName {
+			return i, nil
+		}
+	}
+
+	return 0, errors.Errorf("no %q container found", statefulsets.MySQLServerName)
+}
+
+// splitImage splits an image into its name and version.
+func splitImage(image string) (string, string, error) {
+	parts := strings.Split(image, ":")
+	if len(parts) < 2 {
+		return "", "", errors.Errorf("invalid image %q", image)
+	}
+	return strings.Join(parts[:len(parts)-1], ""), parts[len(parts)-1], nil
+}
+
+func (m *MySQLController) ensureMySQLVersion(c *v1alpha1.Cluster, ss *apps.StatefulSet) error {
+	index, err := getMySQLContainerIndex(ss.Spec.Template.Spec.Containers)
+	if err != nil {
+		return errors.Wrapf(err, "getting MySQL container for StatefulSet %q", ss.Name)
+	}
+	imageName, actualVersion, err := splitImage(ss.Spec.Template.Spec.Containers[index].Image)
+	if err != nil {
+		return errors.Wrapf(err, "getting MySQL version for StatefulSet %q", ss.Name)
+	}
+
+	actual, err := semver.NewVersion(actualVersion)
+	if err != nil {
+		return errors.Wrap(err, "parsing StatuefulSet MySQL version")
+	}
+	expected, err := semver.NewVersion(c.Spec.Version)
+	if err != nil {
+		return errors.Wrap(err, "parsing Cluster MySQL version")
+	}
+
+	switch expected.Compare(*actual) {
+	case -1:
+		return errors.Errorf("attempted unsupported downgrade from %q to %q", actual, expected)
+	case 0:
+		return nil
+	}
+
+	updated := ss.DeepCopy()
+	updated.Spec.Template.Spec.Containers[index].Image = fmt.Sprintf("%s:%s", imageName, c.Spec.Version)
+	// NOTE: We do this as previously we defaulted to the OnDelete strategy
+	// so clusters created with previous versions would not support upgrades.
+	updated.Spec.UpdateStrategy = apps.StatefulSetUpdateStrategy{
+		Type: apps.RollingUpdateStatefulSetStrategyType,
+	}
+
+	err = m.statefulSetControl.Patch(ss, updated)
+	if err != nil {
+		return errors.Wrap(err, "patching StatefulSet")
+	}
+
+	return nil
+}
+
 // ensureMySQLOperatorVersion updates the MySQLOperator resource types that
 //require it to make it consistent with the specified operator version.
 func (m *MySQLController) ensureMySQLOperatorVersion(c *v1alpha1.Cluster, ss *apps.StatefulSet, operatorVersion string) error {

pkg/controllers/cluster/controller_test.go

@@ -30,6 +30,9 @@ import (
 	"k8s.io/client-go/kubernetes/fake"
 	cache "k8s.io/client-go/tools/cache"
 
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
 	"github.com/oracle/mysql-operator/pkg/apis/mysql/v1alpha1"
 	"github.com/oracle/mysql-operator/pkg/constants"
 	"github.com/oracle/mysql-operator/pkg/controllers/util"
@@ -42,6 +45,76 @@ import (
 	buildversion "github.com/oracle/mysql-operator/pkg/version"
 )
 
+func TestGetMySQLContainerIndex(t *testing.T) {
+	testCases := map[string]struct {
+		containers []v1.Container
+		index      int
+		errors     bool
+	}{
+		"empty_errors": {
+			containers: []v1.Container{},
+			errors:     true,
+		},
+		"mysql_server_only": {
+			containers: []v1.Container{{Name: "mysql"}},
+			index:      0,
+		},
+		"mysql_server_and_agent": {
+			containers: []v1.Container{{Name: "mysql-agent"}, {Name: "mysql"}},
+			index:      1,
+		},
+		"mysql_agent_only": {
+			containers: []v1.Container{{Name: "mysql-agent"}},
+			errors:     true,
+		},
+	}
+
+	for name, tc := range testCases {
+		t.Run(name, func(t *testing.T) {
+			index, err := getMySQLContainerIndex(tc.containers)
+			if tc.errors {
+				require.Error(t, err)
+			} else {
+				require.NoError(t, err)
+				assert.Equal(t, index, tc.index)
+			}
+		})
+	}
+}
+
+func TestSplitImage(t *testing.T) {
+	testCases := map[string]struct {
+		image   string
+		name    string
+		version string
+		errors  bool
+	}{
+		"8.0.11": {
+			image:   "mysql/mysql-server:8.0.11",
+			name:    "mysql/mysql-server",
+			version: "8.0.11",
+			errors:  false,
+		},
+		"invalid": {
+			image:  "mysql/mysql-server",
+			errors: true,
+		},
+	}
+
+	for name, tc := range testCases {
+		t.Run(name, func(t *testing.T) {
+			name, version, err := splitImage(tc.image)
+			if tc.errors {
+				require.Error(t, err)
+			} else {
+				require.NoError(t, err)
+				assert.Equal(t, name, tc.name)
+				assert.Equal(t, version, tc.version)
+			}
+		})
+	}
+}
+
 func mockOperatorConfig() operatoropts.MySQLOperatorOpts {
 	opts := operatoropts.MySQLOperatorOpts{}
 	opts.EnsureDefaults()

pkg/resources/statefulsets/statefulset.go

@@ -378,6 +378,9 @@ func NewForCluster(cluster *v1alpha1.Cluster, images operatoropts.Images, servic
 					Volumes:            podVolumes,
 				},
 			},
+			UpdateStrategy: apps.StatefulSetUpdateStrategy{
+				Type: apps.RollingUpdateStatefulSetStrategyType,
+			},
 			ServiceName: serviceName,
 		},
 	}

test/e2e/framework/cluster.go

@@ -19,6 +19,8 @@ import (
 	"strings"
 	"time"
 
+	apps "k8s.io/api/apps/v1beta1"
+	"k8s.io/api/core/v1"
 	apierrors "k8s.io/apimachinery/pkg/api/errors"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/labels"
@@ -35,6 +37,7 @@ import (
 	"github.com/oracle/mysql-operator/pkg/controllers/cluster/labeler"
 	mysqlclientset "github.com/oracle/mysql-operator/pkg/generated/clientset/versioned"
 	"github.com/oracle/mysql-operator/pkg/resources/secrets"
+	"github.com/oracle/mysql-operator/pkg/resources/statefulsets"
 )
 
 // TestDBName is the name of database to use when executing test SQL queries.
@@ -112,7 +115,7 @@ func (j *ClusterTestJig) CreateAndAwaitClusterOrFail(namespace string, members i
 	return cluster
 }
 
-func (j *ClusterTestJig) waitForConditionOrFail(namespace, name string, timeout time.Duration, message string, conditionFn func(*v1alpha1.Cluster) bool) *v1alpha1.Cluster {
+func (j *ClusterTestJig) WaitForConditionOrFail(namespace, name string, timeout time.Duration, message string, conditionFn func(*v1alpha1.Cluster) bool) *v1alpha1.Cluster {
 	var cluster *v1alpha1.Cluster
 	pollFunc := func() (bool, error) {
 		c, err := j.MySQLClient.MySQLV1alpha1().Clusters(namespace).Get(name, metav1.GetOptions{})
@@ -135,12 +138,99 @@ func (j *ClusterTestJig) waitForConditionOrFail(namespace, name string, timeout
 // the running phase.
 func (j *ClusterTestJig) WaitForClusterReadyOrFail(namespace, name string, timeout time.Duration) *v1alpha1.Cluster {
 	Logf("Waiting up to %v for Cluster \"%s/%s\" to be ready", timeout, namespace, name)
-	cluster := j.waitForConditionOrFail(namespace, name, timeout, "have all nodes ready", func(cluster *v1alpha1.Cluster) bool {
+	cluster := j.WaitForConditionOrFail(namespace, name, timeout, "have all nodes ready", func(cluster *v1alpha1.Cluster) bool {
 		return clusterutil.IsClusterReady(cluster)
 	})
 	return cluster
 }
 
+// WaitForClusterUpgradedOrFail waits for a MySQL cluster to be upgraded to the
+// given version or fails.
+func (j *ClusterTestJig) WaitForClusterUpgradedOrFail(namespace, name, version string, timeout time.Duration) *v1alpha1.Cluster {
+	Logf("Waiting up to %v for Cluster \"%s/%s\" to be upgraded", timeout, namespace, name)
+
+	cluster := j.WaitForConditionOrFail(namespace, name, timeout, "be upgraded ", func(cluster *v1alpha1.Cluster) bool {
+		set, err := j.KubeClient.AppsV1beta1().StatefulSets(cluster.Namespace).Get(cluster.Name, metav1.GetOptions{})
+		if err != nil {
+			Failf("Failed to get StatefulSet %[1]q for Cluster %[1]q: %[2]v", name, err)
+		}
+
+		set = j.waitForSSRollingUpdate(set)
+
+		var actualVersion string
+		{
+			var found bool
+			var index int
+			for i, c := range set.Spec.Template.Spec.Containers {
+				if c.Name == statefulsets.MySQLServerName {
+					index = i
+					found = true
+					break
+				}
+			}
+
+			if !found {
+				Failf("no %q container found for StatefulSet %q", statefulsets.MySQLServerName, set.Name)
+			}
+			image := set.Spec.Template.Spec.Containers[index].Image
+			parts := strings.Split(image, ":")
+			if len(parts) < 2 {
+				Failf("invalid image %q for StatefulSet %q", image, set.Name)
+			}
+			actualVersion = parts[len(parts)-1]
+		}
+
+		return actualVersion == version
+	})
+	return cluster
+}
+
+// waitForSSState periodically polls for the ss and its pods until the until function returns either true or an error
+func (j *ClusterTestJig) waitForSSState(ss *apps.StatefulSet, until func(*apps.StatefulSet, *v1.PodList) (bool, error)) {
+	pollErr := wait.PollImmediate(Poll, DefaultTimeout,
+		func() (bool, error) {
+			ssGet, err := j.KubeClient.AppsV1beta1().StatefulSets(ss.Namespace).Get(ss.Name, metav1.GetOptions{})
+			if err != nil {
+				return false, err
+			}
+
+			selector, err := metav1.LabelSelectorAsSelector(ss.Spec.Selector)
+			ExpectNoError(err)
+			podList, err := j.KubeClient.CoreV1().Pods(ss.Namespace).List(metav1.ListOptions{LabelSelector: selector.String()})
+			ExpectNoError(err)
+
+			return until(ssGet, podList)
+		})
+	if pollErr != nil {
+		Failf("Failed waiting for state update: %v", pollErr)
+	}
+}
+
+// waitForRollingUpdate waits for all Pods in set to exist and have the correct revision and for the RollingUpdate to
+// complete. set must have a RollingUpdateStatefulSetStrategyType.
+func (j *ClusterTestJig) waitForSSRollingUpdate(set *apps.StatefulSet) *apps.StatefulSet {
+	var pods *v1.PodList
+	if set.Spec.UpdateStrategy.Type != apps.RollingUpdateStatefulSetStrategyType {
+		Failf("StatefulSet %s/%s attempt to wait for rolling update with updateStrategy %s",
+			set.Namespace,
+			set.Name,
+			set.Spec.UpdateStrategy.Type)
+	}
+	Logf("Waiting for StatefulSet %s/%s to complete update", set.Namespace, set.Name)
+	j.waitForSSState(set, func(set2 *apps.StatefulSet, pods2 *v1.PodList) (bool, error) {
+		set = set2
+		pods = pods2
+		if len(pods.Items) < int(*set.Spec.Replicas) {
+			return false, nil
+		}
+		if set.Status.UpdateRevision != set.Status.CurrentRevision {
+			return false, nil
+		}
+		return true, nil
+	})
+	return set
+}
+
 // SanityCheckCluster checks basic properties of a given Cluster match
 // our expectations.
 func (j *ClusterTestJig) SanityCheckCluster(cluster *v1alpha1.Cluster) {

test/e2e/upgrade.go

@@ -0,0 +1,62 @@
+// Copyright 2018 Oracle and/or its affiliates. All rights reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+package e2e
+
+import (
+	. "github.com/onsi/ginkgo"
+	. "github.com/onsi/gomega"
+
+	"github.com/oracle/mysql-operator/pkg/apis/mysql/v1alpha1"
+	mysqlclientset "github.com/oracle/mysql-operator/pkg/generated/clientset/versioned"
+	"github.com/oracle/mysql-operator/test/e2e/framework"
+)
+
+var _ = Describe("MySQL Upgrade", func() {
+	f := framework.NewDefaultFramework("upgrade")
+
+	var mcs mysqlclientset.Interface
+	BeforeEach(func() {
+		mcs = f.MySQLClientSet
+	})
+
+	It("should be possible to upgrade a cluster from 8.0.11 to 8.0.12", func() {
+		jig := framework.NewClusterTestJig(mcs, f.ClientSet, "upgrade-test")
+
+		By("creating an 8.0.11 cluster")
+
+		cluster := jig.CreateAndAwaitClusterOrFail(f.Namespace.Name, 3, func(c *v1alpha1.Cluster) {
+			c.Spec.Version = "8.0.11"
+		}, framework.DefaultTimeout)
+
+		expected, err := framework.WriteSQLTest(cluster, cluster.Name+"-0")
+		Expect(err).NotTo(HaveOccurred())
+
+		By("triggering an upgrade to 8.0.12")
+
+		cluster.Spec.Version = "8.0.12"
+		cluster, err = mcs.MySQLV1alpha1().Clusters(cluster.Namespace).Update(cluster)
+		Expect(err).NotTo(HaveOccurred())
+
+		By("waiting for the upgrade to complete")
+
+		cluster = jig.WaitForClusterUpgradedOrFail(cluster.Namespace, cluster.Name, "8.0.12", framework.DefaultTimeout)
+
+		By("testing we can read from the upgraded database")
+
+		actual, err := framework.ReadSQLTest(cluster, cluster.Name+"-0")
+		Expect(err).NotTo(HaveOccurred())
+		Expect(actual).To(Equal(expected))
+	})
+})