[GitHub] [hbase] joshelser commented on a diff in pull request #4979: HBASE-27574 Implement ClusterManager interface for Kubernetes

[GitBox <gi...@apache.org>]

joshelser commented on code in PR #4979:
URL: https://github.com/apache/hbase/pull/4979#discussion_r1081422560


##########
hbase-it/src/test/java/org/apache/hadoop/hbase/KubernetesClusterManager.java:
##########
@@ -0,0 +1,258 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you 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 org.apache.hadoop.hbase;
+
+import com.google.gson.JsonSyntaxException;
+import io.kubernetes.client.openapi.ApiClient;
+import io.kubernetes.client.openapi.ApiException;
+import io.kubernetes.client.openapi.apis.CoreV1Api;
+import io.kubernetes.client.openapi.models.V1ObjectMeta;
+import io.kubernetes.client.openapi.models.V1Pod;
+import io.kubernetes.client.openapi.models.V1PodList;
+import io.kubernetes.client.util.ClientBuilder;
+import java.io.IOException;
+import java.util.Collections;
+import java.util.HashSet;
+import java.util.Optional;
+import java.util.Set;
+import org.apache.hadoop.conf.Configuration;
+import org.apache.hadoop.conf.Configured;
+import org.apache.hadoop.hbase.procedure2.util.StringUtils;
+import org.apache.yetus.audience.InterfaceAudience;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Cluster manager for K8s context. Supports taking destructive actions, and checking for the
+ * presence of a running process. Assumes services are running behind some type of `Deployment` that
+ * will handle starting replacement processes after a destructive action. Requires that the
+ * configuration specify a Kubernetes namespace.
+ * </p>
+ * Note that the k8s java client is a bit dodgy unable to read responses when we call delete (i.e.
+ * kill) and sometimes when checking isRunning; makes for noisy logs and sometimes the operations
+ * overrun each other but generally succeed.
+ */
+@InterfaceAudience.LimitedPrivate(HBaseInterfaceAudience.CONFIG)
+public class KubernetesClusterManager extends Configured implements ClusterManager {
+  private static final Logger LOG = LoggerFactory.getLogger(KubernetesClusterManager.class);
+
+  private static final String NAMESPACE_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".namespace";
+  private static final String ZOOKEEPER_ROLE_SELECTOR_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".zookeeper_role";
+  private static final String NAMENODE_ROLE_SELECTOR_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".namenode_role";
+  private static final String JOURNALNODE_ROLE_SELECTOR_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".journalnode_role";
+  private static final String DATANODE_ROLE_SELECTOR_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".datanode_role";
+  private static final String MASTER_ROLE_SELECTOR_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".master_role";
+  private static final String REGIONSERVER_ROLE_SELECTOR_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".regionserver_role";
+  private static final Set<ServiceType> SUPPORTED_SERVICE_TYPES = buildSupportedServiceTypesSet();
+
+  private CoreV1Api api;
+  private String namespace;
+
+  private static Set<ServiceType> buildSupportedServiceTypesSet() {
+    final Set<ServiceType> set = new HashSet<>();
+    set.add(ServiceType.ZOOKEEPER_SERVER);
+    set.add(ServiceType.HADOOP_NAMENODE);
+    set.add(ServiceType.HADOOP_JOURNALNODE);
+    set.add(ServiceType.HADOOP_DATANODE);
+    set.add(ServiceType.HBASE_MASTER);
+    set.add(ServiceType.HBASE_REGIONSERVER);
+    return Collections.unmodifiableSet(set);
+  }
+
+  @Override
+  public void setConf(Configuration configuration) {
+    // This is usually called from the constructor (The call to super goes to
+    // Configured which does a setConf in its construction) but the configuration
+    // null at that time. After construction, setConf is called again.
+    // Assume single-thread doing ClusterManager setup.
+    if (configuration == null) {
+      LOG.debug("Skipping because provided configuration=null");
+      return;
+    }
+    if (getConf() != null) {
+      LOG.debug("Skipping because configuration already set, getConf={}", getConf());
+      return;
+    }
+    super.setConf(configuration);
+    this.namespace = configuration.get(NAMESPACE_CONF_KEY);
+    LOG.info(
+      "Configuration={}, namespace={}, hbase.rootdir={}, hbase.zookeeper.quorum={}, "
+        + "hbase.client.zookeeper.quorum={}",
+      configuration, this.namespace, configuration.get("hbase.rootdir"),
+      configuration.get("hbase.zookeeper.quorum"),
+      configuration.get("hbase.client.zookeeper.quorum"));
+    final CoreV1Api coreV1Api;
+    try {
+      ApiClient client = ClientBuilder.cluster().build();
+      io.kubernetes.client.openapi.Configuration.setDefaultApiClient(client);
+      coreV1Api = new CoreV1Api();
+    } catch (IOException ioe) {
+      throw new RuntimeException("Failed Kubernetes ApiClient construction", ioe);
+    }
+    this.api = coreV1Api;
+  }
+
+  @Override
+  public void start(ServiceType service, String hostname, int port) {
+    assertSupportedServiceType(service);
+    // presumably kubernetes is automatically starting pods, so nothing to do here.
+  }
+
+  @Override
+  public void stop(ServiceType service, String hostname, int port) throws IOException {
+    assertSupportedServiceType(service);
+    kill(hostname);
+  }
+
+  @Override
+  public void restart(ServiceType service, String hostname, int port) throws IOException {
+    assertSupportedServiceType(service);
+    kill(hostname);
+  }
+
+  @Override
+  public void kill(ServiceType service, String hostname, int port) throws IOException {
+    assertSupportedServiceType(service);
+    kill(hostname);
+  }
+
+  @Override
+  public void suspend(ServiceType service, String hostname, int port) {
+    throw unsupportedActionType("suspend");
+  }
+
+  @Override
+  public void resume(ServiceType service, String hostname, int port) {
+    throw unsupportedActionType("resume");
+  }
+
+  @Override
+  public boolean isRunning(ServiceType service, String hostname, int port) throws IOException {
+    final String roleSelector = roleSelectorForServiceType(service);
+    return isRunning(roleSelector, hostname);
+  }
+
+  private String roleSelectorForServiceType(final ServiceType serviceType) {
+    assertSupportedServiceType(serviceType);
+    switch (serviceType) {
+      case HADOOP_NAMENODE:
+        return getConf().get(NAMENODE_ROLE_SELECTOR_CONF_KEY, "namenode");
+      case HADOOP_DATANODE:
+        return getConf().get(DATANODE_ROLE_SELECTOR_CONF_KEY, "datanode");
+      case HADOOP_JOURNALNODE:
+        return getConf().get(JOURNALNODE_ROLE_SELECTOR_CONF_KEY, "journalnode");
+      case ZOOKEEPER_SERVER:
+        return getConf().get(ZOOKEEPER_ROLE_SELECTOR_CONF_KEY, "zookeeper");
+      case HBASE_MASTER:
+        return getConf().get(MASTER_ROLE_SELECTOR_CONF_KEY, "master");
+      case HBASE_REGIONSERVER:
+        return getConf().get(REGIONSERVER_ROLE_SELECTOR_CONF_KEY, "regionserver");
+      default:
+        throw new RuntimeException("should not happen");
+    }
+  }
+
+  private boolean isRunning(String roleSelector, String hostname) throws IOException {
+    if (api == null) {
+      return false;
+    }
+    final V1PodList list;
+    try {
+      list = api.listNamespacedPod(namespace, null, null, null, null,
+        "role in (" + roleSelector + ")", null, null, null, null, null);
+    } catch (ApiException e) {
+      throw convertApiException(e);
+    }
+    if (list == null) {
+      return false;
+    }
+    for (V1Pod item : list.getItems()) {
+      final String podName =
+        Optional.ofNullable(item).map(V1Pod::getMetadata).map(V1ObjectMeta::getName).orElse(null);
+      if (StringUtils.isEmpty(podName)) {
+        LOG.warn("Listing of namespace '{}' contains entry with empty pod name.", namespace);
+        continue;
+      }
+      if (hostname.startsWith(podName)) {
+        return true;
+      }

Review Comment:
   Could you also check the Status and Readiness probes for the pod, rather than just existence of the pod?
   
   I would assume you have some kind of readiness probe defined (e.g. one to check that the server is listening on its port is pretty easy to do) for the HBase processes.



##########
hbase-it/src/test/java/org/apache/hadoop/hbase/KubernetesClusterManager.java:
##########
@@ -0,0 +1,258 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you 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 org.apache.hadoop.hbase;
+
+import com.google.gson.JsonSyntaxException;
+import io.kubernetes.client.openapi.ApiClient;
+import io.kubernetes.client.openapi.ApiException;
+import io.kubernetes.client.openapi.apis.CoreV1Api;
+import io.kubernetes.client.openapi.models.V1ObjectMeta;
+import io.kubernetes.client.openapi.models.V1Pod;
+import io.kubernetes.client.openapi.models.V1PodList;
+import io.kubernetes.client.util.ClientBuilder;
+import java.io.IOException;
+import java.util.Collections;
+import java.util.HashSet;
+import java.util.Optional;
+import java.util.Set;
+import org.apache.hadoop.conf.Configuration;
+import org.apache.hadoop.conf.Configured;
+import org.apache.hadoop.hbase.procedure2.util.StringUtils;
+import org.apache.yetus.audience.InterfaceAudience;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * Cluster manager for K8s context. Supports taking destructive actions, and checking for the
+ * presence of a running process. Assumes services are running behind some type of `Deployment` that
+ * will handle starting replacement processes after a destructive action. Requires that the
+ * configuration specify a Kubernetes namespace.
+ * </p>
+ * Note that the k8s java client is a bit dodgy unable to read responses when we call delete (i.e.
+ * kill) and sometimes when checking isRunning; makes for noisy logs and sometimes the operations
+ * overrun each other but generally succeed.
+ */
+@InterfaceAudience.LimitedPrivate(HBaseInterfaceAudience.CONFIG)
+public class KubernetesClusterManager extends Configured implements ClusterManager {
+  private static final Logger LOG = LoggerFactory.getLogger(KubernetesClusterManager.class);
+
+  private static final String NAMESPACE_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".namespace";
+  private static final String ZOOKEEPER_ROLE_SELECTOR_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".zookeeper_role";
+  private static final String NAMENODE_ROLE_SELECTOR_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".namenode_role";
+  private static final String JOURNALNODE_ROLE_SELECTOR_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".journalnode_role";
+  private static final String DATANODE_ROLE_SELECTOR_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".datanode_role";
+  private static final String MASTER_ROLE_SELECTOR_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".master_role";
+  private static final String REGIONSERVER_ROLE_SELECTOR_CONF_KEY =
+    KubernetesClusterManager.class.getCanonicalName() + ".regionserver_role";
+  private static final Set<ServiceType> SUPPORTED_SERVICE_TYPES = buildSupportedServiceTypesSet();
+
+  private CoreV1Api api;
+  private String namespace;
+
+  private static Set<ServiceType> buildSupportedServiceTypesSet() {
+    final Set<ServiceType> set = new HashSet<>();
+    set.add(ServiceType.ZOOKEEPER_SERVER);
+    set.add(ServiceType.HADOOP_NAMENODE);
+    set.add(ServiceType.HADOOP_JOURNALNODE);
+    set.add(ServiceType.HADOOP_DATANODE);
+    set.add(ServiceType.HBASE_MASTER);
+    set.add(ServiceType.HBASE_REGIONSERVER);
+    return Collections.unmodifiableSet(set);
+  }
+
+  @Override
+  public void setConf(Configuration configuration) {
+    // This is usually called from the constructor (The call to super goes to
+    // Configured which does a setConf in its construction) but the configuration
+    // null at that time. After construction, setConf is called again.
+    // Assume single-thread doing ClusterManager setup.
+    if (configuration == null) {
+      LOG.debug("Skipping because provided configuration=null");
+      return;
+    }
+    if (getConf() != null) {
+      LOG.debug("Skipping because configuration already set, getConf={}", getConf());
+      return;
+    }
+    super.setConf(configuration);
+    this.namespace = configuration.get(NAMESPACE_CONF_KEY);

Review Comment:
   Use the two-arg `get()` method to give a default value?



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: issues-unsubscribe@hbase.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org