Support specifying Service type

Author: sbernauer

CHANGELOG.md

@@ -12,8 +12,13 @@ All notable changes to this project will be documented in this file.
 
 ### Changed
 
-- `operator-rs` `0.36.0` -> `0.39.0` ([#326], [#337]).
 - [Breaking] Moved top level config option to `clusterConfig` ([#326]).
+- [BREAKING] Support specifying Service type.
+  This enables us to later switch non-breaking to using `ListenerClasses` for the exposure of Services.
+  This change is breaking, because - for security reasons - we default to the `cluster-internal` `ListenerClass`.
+  If you need you cluster to be accessible from outside of Kubernetes you need to set `clusterConfig.listenerClass`
+  to `external-unstable` ([#XXX]).
+  - `operator-rs` `0.36.0` -> `0.39.0` ([#326], [#337]).
 
 ### Removed
 

rust/crd/src/lib.rs

@@ -90,6 +90,39 @@ pub struct HdfsClusterConfig {
     pub vector_aggregator_config_map_name: Option<String>,
     /// Name of the ZooKeeper discovery config map.
     pub zookeeper_config_map_name: String,
+    /// In the future this setting will control, which ListenerClass <https://docs.stackable.tech/home/stable/listener-operator/listenerclass.html>
+    /// will be used to expose the service.
+    /// Currently only a subset of the ListenerClasses are supported by choosing the type of the created Services
+    /// by looking at the ListenerClass name specified,
+    /// In a future release support for custom ListenerClasses will be introduced without a breaking change:
+    ///
+    /// * cluster-internal: Use a ClusterIP service
+    ///
+    /// * external-unstable: Use a NodePort service
+    #[serde(default)]
+    pub listener_class: CurrentlySupportedListenerClasses,
+}
+
+// TODO: Temporary solution until listener-operator is finished
+#[derive(
+    Clone, Debug, Default, Display, Deserialize, Eq, Hash, JsonSchema, PartialEq, Serialize,
+)]
+#[serde(rename_all = "PascalCase")]
+pub enum CurrentlySupportedListenerClasses {
+    #[default]
+    #[serde(rename = "cluster-internal")]
+    ClusterInternal,
+    #[serde(rename = "external-unstable")]
+    ExternalUnstable,
+}
+
+impl CurrentlySupportedListenerClasses {
+    pub fn k8s_service_type(&self) -> String {
+        match self {
+            CurrentlySupportedListenerClasses::ClusterInternal => "ClusterIP".to_string(),
+            CurrentlySupportedListenerClasses::ExternalUnstable => "NodePort".to_string(),
+        }
+    }
 }
 
 /// This is a shared trait for all role/role-group config structs to avoid duplication
@@ -433,9 +466,15 @@ impl HdfsCluster {
         );
         group_labels.insert(String::from("role"), rolegroup_ref.role.clone());
         group_labels.insert(String::from("group"), rolegroup_ref.role_group.clone());
-        // TODO: in a production environment, probably not all roles need to be exposed with one NodePort per Pod but it's
-        // useful for development purposes.
-        group_labels.insert(LABEL_ENABLE.to_string(), "true".to_string());
+
+        if self.spec.cluster_config.listener_class
+            == CurrentlySupportedListenerClasses::ExternalUnstable
+        {
+            // TODO: in a production environment, probably not all roles need to be exposed with one NodePort per Pod but it's
+            // useful for development purposes.
+
+            group_labels.insert(LABEL_ENABLE.to_string(), "true".to_string());
+        }
 
         group_labels
     }

rust/operator/src/hdfs_controller.rs

@@ -338,6 +338,8 @@ fn rolegroup_service(
             .with_label("prometheus.io/scrape", "true")
             .build(),
         spec: Some(ServiceSpec {
+            // Internal communication does not need to be exposed
+            type_: Some("ClusterIP".to_string()),
             cluster_ip: Some("None".to_string()),
             ports: Some(
                 role.ports()

tests/templates/kuttl/smoke/01-install-zk.yaml.j2

@@ -7,8 +7,9 @@ spec:
   image:
     productVersion: "{{ test_scenario['values']['zookeeper'].split('-stackable')[0] }}"
     stackableVersion: "{{ test_scenario['values']['zookeeper'].split('-stackable')[1] }}"
-{% if lookup('env', 'VECTOR_AGGREGATOR') %}
   clusterConfig:
+    listenerClass: {{ test_scenario['values']['listener-class'] }}
+{% if lookup('env', 'VECTOR_AGGREGATOR') %}
     logging:
       vectorAggregatorConfigMapName: vector-aggregator-discovery
 {% endif %}

tests/templates/kuttl/smoke/02-install-hdfs.yaml.j2

@@ -16,6 +16,7 @@ spec:
   clusterConfig:
     dfsReplication: 1
     zookeeperConfigMapName: hdfs-zk
+    listenerClass: {{ test_scenario['values']['listener-class'] }}
 {% if lookup('env', 'VECTOR_AGGREGATOR') %}
     vectorAggregatorConfigMapName: vector-aggregator-discovery
 {% endif %}

tests/templates/kuttl/smoke/04-assert.yaml

@@ -2,4 +2,4 @@
 apiVersion: kuttl.dev/v1beta1
 kind: TestAssert
 commands:
-  - script: kubectl exec -n $NAMESPACE webhdfs-0 -- python /tmp/webhdfs.py ls
+  - script: kubectl exec -n $NAMESPACE webhdfs-0 -- python /tmp/webhdfs.py $NAMESPACE ls

tests/templates/kuttl/smoke/04-create-file.yaml

@@ -4,4 +4,4 @@ kind: TestStep
 commands:
   - script: kubectl cp -n $NAMESPACE ./webhdfs.py  webhdfs-0:/tmp
   - script: kubectl cp -n $NAMESPACE ./testdata.txt webhdfs-0:/tmp
-  - script: kubectl exec -n $NAMESPACE webhdfs-0 -- python /tmp/webhdfs.py create
+  - script: kubectl exec -n $NAMESPACE webhdfs-0 -- python /tmp/webhdfs.py $NAMESPACE create

tests/templates/kuttl/smoke/webhdfs.py

@@ -5,7 +5,8 @@
 
 def main() -> int:
     result = 0
-    command = sys.argv[1]
+    namespace = sys.argv[1]
+    command = sys.argv[2]
 
     log_level = "DEBUG"
     logging.basicConfig(
@@ -16,7 +17,7 @@ def main() -> int:
 
     if command == "ls":
         http_code = requests.get(
-            "http://hdfs-namenode-default-0:9870/webhdfs/v1/testdata.txt?user.name=stackable&op=LISTSTATUS"
+            f"http://hdfs-namenode-default-0.hdfs-namenode-default.{namespace}.svc.cluster.local:9870/webhdfs/v1/testdata.txt?user.name=stackable&op=LISTSTATUS"
         ).status_code
         if http_code != 200:
             result = 1
@@ -30,7 +31,7 @@ def main() -> int:
             )
         }
         http_code = requests.put(
-            "http://hdfs-namenode-default-0:9870/webhdfs/v1/testdata.txt?user.name=stackable&op=CREATE",
+            f"http://hdfs-namenode-default-0.hdfs-namenode-default.{namespace}.svc.cluster.local:9870/webhdfs/v1/testdata.txt?user.name=stackable&op=CREATE",
             files=files,
             allow_redirects=True,
         ).status_code

tests/test-definition.yaml

@@ -22,13 +22,19 @@ dimensions:
     values:
       - "default"
       - "2hdd-1ssd"
+  # Used for both, zookeeper and hdfs
+  - name: listener-class
+    values:
+      - "cluster-internal"
+      - "external-unstable"
 tests:
   - name: smoke
     dimensions:
       - hadoop
       - zookeeper
       - number-of-datanodes
       - datanode-pvcs
+      - listener-class
   - name: orphaned-resources
     dimensions:
       - hadoop-latest