Add support for Cassandra create-schema job

CONTRIBUTING.adoc

@@ -111,9 +111,10 @@ Accessing the provided "address" in your web browser should display the Jaeger U
 
 ==== Storage configuration
 
-There's a template under the `test` directory that can be used to setup an Elasticsearch cluster. Alternatively, the following command can be executed to install it:
+There are a set of templates under the `test` directory that can be used to setup an Elasticsearch and/or Cassandra cluster. Alternatively, the following commands can be executed to install it:
 
 [source,bash]
 ----
 make es
+make cassandra
 ----

Makefile

@@ -53,7 +53,7 @@ unit-tests:
 	@go test $(PACKAGES) -cover -coverprofile=cover.out
 
 .PHONY: e2e-tests
-e2e-tests: es crd build docker push
+e2e-tests: cassandra es crd build docker push
 	@echo Running end-to-end tests...
 	@cp deploy/rbac.yaml deploy/test/namespace-manifests.yaml
 	@echo "---" >> deploy/test/namespace-manifests.yaml
@@ -68,6 +68,10 @@ run: crd
 es:
 	@kubectl create -f ./test/elasticsearch.yml 2>&1 | grep -v "already exists" || true
 
+.PHONY: cassandra
+cassandra:
+	@kubectl create -f ./test/cassandra.yml 2>&1 | grep -v "already exists" || true
+
 .PHONY: crd
 crd:
 	@kubectl create -f deploy/crd.yaml 2>&1 | grep -v "already exists" || true

README.adoc

@@ -242,6 +242,53 @@ spec:
               fieldPath: status.hostIP
 ----
 
+== Schema migration
+
+=== Cassandra
+
+When the storage type is set to Cassandra, the operator will automatically create a batch job that creates the required schema for Jaeger to run. This batch job will block the Jaeger installation, so that it starts only after the schema is successfuly created. The creation of this batch job can be disabled by setting the `enabled` property to `false`:
+
+[source,yaml]
+----
+apiVersion: io.jaegertracing/v1alpha1
+kind: Jaeger
+metadata:
+  name: cassandra-without-create-schema
+spec:
+  strategy: all-in-one
+  storage:
+    type: cassandra
+    cassandra-create-schema:
+      enabled: false # <1>
+----
+<1> Defaults to `true`
+
+Further aspects of the batch job can be configured as well. An example with all the possible options is shown below:
+
+[source,yaml]
+----
+apiVersion: io.jaegertracing/v1alpha1
+kind: Jaeger
+metadata:
+  name: cassandra-with-create-schema
+spec:
+  strategy: all-in-one # <1>
+  storage:
+    type: cassandra
+    options: # <2>
+      cassandra:
+        servers: cassandra
+        keyspace: jaeger_v1_datacenter3
+    cassandra-create-schema: # <3>
+      datacenter: "datacenter3"
+      mode: "test"
+----
+<1> The same works for `production`
+<2> These options are for the regular Jaeger components, like `collector` and `query`
+<3> The options for the `create-schema` job
+
+NOTE: the default create-schema job uses `MODE=prod`, which implies a replication factor of `2`, using `NetworkTopologyStrategy` as the class, effectively meaning that at least 3 nodes are required in the Cassandra cluster. If a `SimpleStrategy` is desired, set the mode to `test`, which then sets the replication factor of `1`. Refer to the link:https://github.com/jaegertracing/jaeger/blob/v1.7.0/plugin/storage/cassandra/schema/create.sh[create-schema script] for more details.
+
 == Removing an instance
 
 To remove an instance, just use the `delete` command with the file used for the instance creation:

deploy/examples/with-cassandra.yaml

@@ -0,0 +1,15 @@
+apiVersion: io.jaegertracing/v1alpha1
+kind: Jaeger
+metadata:
+  name: with-cassandra
+spec:
+  strategy: all-in-one
+  storage:
+    type: cassandra
+    options:
+      cassandra:
+        servers: cassandra
+        keyspace: jaeger_v1_datacenter3
+    cassandra-create-schema:
+      datacenter: "datacenter3"
+      mode: "test"

deploy/rbac.yaml

@@ -36,7 +36,12 @@ rules:
   - ingresses
   verbs:
   - "*"
-
+- apiGroups:
+  - batch
+  resources:
+  - jobs
+  verbs:
+  - "*"
 ---
 
 kind: RoleBinding

pkg/apis/io/v1alpha1/options.go

@@ -68,3 +68,9 @@ func (o *Options) ToArgs() []string {
 
 	return nil
 }
+
+// Map returns a map representing the option entries. Items are flattened, with dots as separators. For instance
+// an option "cassandra" with a nested "servers" object becomes an entry with the key "cassandra.servers"
+func (o *Options) Map() map[string]string {
+	return o.opts
+}

pkg/apis/io/v1alpha1/options_test.go

@@ -52,3 +52,9 @@ func TestMultipleSubValues(t *testing.T) {
 	args := o.ToArgs()
 	assert.Len(t, args, 3)
 }
+
+func TestExposedMap(t *testing.T) {
+	o := NewOptions(nil)
+	o.UnmarshalJSON([]byte(`{"cassandra": {"servers": "cassandra:9042"}}`))
+	assert.Equal(t, "cassandra:9042", o.Map()["cassandra.servers"])
+}

pkg/apis/io/v1alpha1/types.go

@@ -74,6 +74,15 @@ type JaegerAgentSpec struct {
 
 // JaegerStorageSpec defines the common storage options to be used for the query and collector
 type JaegerStorageSpec struct {
-	Type    string  `json:"type"` // can be `memory` (default), `cassandra`, `elasticsearch`, `kafka` or `managed`
-	Options Options `json:"options"`
+	Type                  string                          `json:"type"` // can be `memory` (default), `cassandra`, `elasticsearch`, `kafka` or `managed`
+	Options               Options                         `json:"options"`
+	CassandraCreateSchema JaegerCassandraCreateSchemaSpec `json:"cassandra-create-schema"`
+}
+
+// JaegerCassandraCreateSchemaSpec holds the options related to the create-schema batch job
+type JaegerCassandraCreateSchemaSpec struct {
+	Enabled    *bool  `json:"enabled"`
+	Image      string `json:"image"`
+	Datacenter string `json:"datacenter"`
+	Mode       string `json:"mode"`
 }

pkg/cmd/start/main.go

@@ -44,6 +44,9 @@ func NewStartCommand() *cobra.Command {
 	cmd.Flags().StringP("jaeger-all-in-one-image", "", "jaegertracing/all-in-one", "The Docker image for the Jaeger all-in-one")
 	viper.BindPFlag("jaeger-all-in-one-image", cmd.Flags().Lookup("jaeger-all-in-one-image"))
 
+	cmd.Flags().StringP("jaeger-cassandra-schema-image", "", "jaegertracing/jaeger-cassandra-schema", "The Docker image for the Jaeger Cassandra Schema")
+	viper.BindPFlag("jaeger-cassandra-schema-image", cmd.Flags().Lookup("jaeger-cassandra-schema-image"))
+
 	return cmd
 }
 

pkg/controller/all-in-one.go

@@ -5,9 +5,11 @@ import (
 
 	"github.com/operator-framework/operator-sdk/pkg/sdk"
 	"github.com/sirupsen/logrus"
+	batchv1 "k8s.io/api/batch/v1"
 
 	"github.com/jaegertracing/jaeger-operator/pkg/apis/io/v1alpha1"
 	"github.com/jaegertracing/jaeger-operator/pkg/deployment"
+	"github.com/jaegertracing/jaeger-operator/pkg/storage"
 )
 
 type allInOneController struct {
@@ -22,7 +24,7 @@ func newAllInOneController(ctx context.Context, jaeger *v1alpha1.Jaeger) *allInO
 	}
 }
 
-func (c allInOneController) Create() []sdk.Object {
+func (c *allInOneController) Create() []sdk.Object {
 	logrus.Debugf("Creating all-in-one for '%v'", c.jaeger.Name)
 
 	dep := deployment.NewAllInOne(c.jaeger)
@@ -43,7 +45,11 @@ func (c allInOneController) Create() []sdk.Object {
 	return os
 }
 
-func (c allInOneController) Update() []sdk.Object {
+func (c *allInOneController) Update() []sdk.Object {
 	logrus.Debug("Update isn't available for all-in-one")
 	return []sdk.Object{}
 }
+
+func (c *allInOneController) Dependencies() []batchv1.Job {
+	return storage.Dependencies(c.jaeger)
+}

pkg/controller/all-in-one_test.go

@@ -10,6 +10,7 @@ import (
 	"github.com/stretchr/testify/assert"
 
 	"github.com/jaegertracing/jaeger-operator/pkg/apis/io/v1alpha1"
+	"github.com/jaegertracing/jaeger-operator/pkg/storage"
 )
 
 func init() {
@@ -41,6 +42,12 @@ func TestUpdateAllInOneDeployment(t *testing.T) {
 	assert.Len(t, objs, 0)
 }
 
+func TestDelegateAllInOneDepedencies(t *testing.T) {
+	// for now, we just have storage dependencies
+	c := newAllInOneController(context.TODO(), v1alpha1.NewJaeger("TestDelegateAllInOneDepedencies"))
+	assert.Equal(t, c.Dependencies(), storage.Dependencies(c.jaeger))
+}
+
 func assertDeploymentsAndServicesForAllInOne(t *testing.T, name string, objs []sdk.Object, hasDaemonSet bool) {
 	if hasDaemonSet {
 		assert.Len(t, objs, 7)

pkg/controller/controller.go

@@ -6,12 +6,14 @@ import (
 
 	"github.com/operator-framework/operator-sdk/pkg/sdk"
 	"github.com/sirupsen/logrus"
+	batchv1 "k8s.io/api/batch/v1"
 
 	"github.com/jaegertracing/jaeger-operator/pkg/apis/io/v1alpha1"
 )
 
 // Controller knows what type of deployments to build based on a given spec
 type Controller interface {
+	Dependencies() []batchv1.Job
 	Create() []sdk.Object
 	Update() []sdk.Object
 }

pkg/controller/production.go

@@ -5,9 +5,11 @@ import (
 
 	"github.com/operator-framework/operator-sdk/pkg/sdk"
 	"github.com/sirupsen/logrus"
+	batchv1 "k8s.io/api/batch/v1"
 
 	"github.com/jaegertracing/jaeger-operator/pkg/apis/io/v1alpha1"
 	"github.com/jaegertracing/jaeger-operator/pkg/deployment"
+	"github.com/jaegertracing/jaeger-operator/pkg/storage"
 )
 
 type productionController struct {
@@ -56,3 +58,7 @@ func (c *productionController) Update() []sdk.Object {
 	logrus.Debug("Update isn't yet available")
 	return []sdk.Object{}
 }
+
+func (c *productionController) Dependencies() []batchv1.Job {
+	return storage.Dependencies(c.jaeger)
+}

pkg/controller/production_test.go

@@ -11,6 +11,7 @@ import (
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 
 	"github.com/jaegertracing/jaeger-operator/pkg/apis/io/v1alpha1"
+	"github.com/jaegertracing/jaeger-operator/pkg/storage"
 )
 
 func init() {
@@ -77,6 +78,12 @@ func TestOptionsArePassed(t *testing.T) {
 	}
 }
 
+func TestDelegateProductionDepedencies(t *testing.T) {
+	// for now, we just have storage dependencies
+	c := newProductionController(context.TODO(), v1alpha1.NewJaeger("TestDelegateProductionDepedencies"))
+	assert.Equal(t, c.Dependencies(), storage.Dependencies(c.jaeger))
+}
+
 func assertDeploymentsAndServicesForProduction(t *testing.T, name string, objs []sdk.Object, hasDaemonSet bool) {
 	if hasDaemonSet {
 		assert.Len(t, objs, 7)