Update pauser to handle exception and run unpause as long as possible

lib/build.gradle

@@ -8,6 +8,7 @@ dependencies {
     testImplementation(platform("org.junit:junit-bom:${junitVersion}"))
     testImplementation 'org.junit.jupiter:junit-jupiter'
     testImplementation "org.mockito:mockito-core:${mockitoVersion}"
+    testImplementation "org.mockito:mockito-inline:${mockitoVersion}"
 }
 
 test {

lib/src/main/java/com/scalar/admin/kubernetes/Pauser.java

@@ -1,5 +1,6 @@
 package com.scalar.admin.kubernetes;
 
+import com.google.common.annotations.VisibleForTesting;
 import com.google.common.util.concurrent.Uninterruptibles;
 import com.scalar.admin.RequestCoordinator;
 import io.kubernetes.client.openapi.Configuration;
@@ -34,7 +35,7 @@
 @NotThreadSafe
 public class Pauser {
 
-  private static final int MAX_UNPAUSE_RETRY_COUNT = 3;
+  @VisibleForTesting static final int MAX_UNPAUSE_RETRY_COUNT = 3;
 
   private final Logger logger = LoggerFactory.getLogger(Pauser.class);
   private final TargetSelector targetSelector;
@@ -79,26 +80,50 @@ public PausedDuration pause(int pauseDuration, @Nullable Long maxPauseWaitTime)
 
     TargetSnapshot target;
     try {
-      target = targetSelector.select();
+      target = getTarget();
     } catch (Exception e) {
       throw new PauserException("Failed to find the target pods to pause.", e);
     }
 
-    RequestCoordinator coordinator = getRequestCoordinator(target);
+    RequestCoordinator coordinator;
+    try {
+      coordinator = getRequestCoordinator(target);
+    } catch (Exception e) {
+      throw new PauserException("Failed to initialize the coordinator.", e);
+    }
 
-    coordinator.pause(true, maxPauseWaitTime);
+    Instant startTime;
+    Instant endTime;
+    try {
+      coordinator.pause(true, maxPauseWaitTime);
 
-    Instant startTime = Instant.now();
+      startTime = Instant.now();
 
-    Uninterruptibles.sleepUninterruptibly(pauseDuration, TimeUnit.MILLISECONDS);
+      Uninterruptibles.sleepUninterruptibly(pauseDuration, TimeUnit.MILLISECONDS);
 
-    Instant endTime = Instant.now();
+      endTime = Instant.now();
 
-    unpauseWithRetry(coordinator, MAX_UNPAUSE_RETRY_COUNT, target);
+      unpauseWithRetry(coordinator, MAX_UNPAUSE_RETRY_COUNT, target);
+
+    } catch (Exception e) {
+      try {
+        unpauseWithRetry(coordinator, MAX_UNPAUSE_RETRY_COUNT, target);
+      } catch (PauserException ex) {
+        throw new PauserException("unpauseWithRetry() method failed twice.", e);
+      } catch (Exception ex) {
+        throw new PauserException(
+            "unpauseWithRetry() method failed twice due to unexpected exception.", e);
+      }
+      throw new PauserException(
+          "The pause operation failed for some reason. However, the unpause operation succeeded"
+              + " afterward. Currently, the scalar products are running with the unpause status."
+              + " You should retry the pause operation to ensure proper backup.",
+          e);
+    }
 
     TargetSnapshot targetAfterPause;
     try {
-      targetAfterPause = targetSelector.select();
+      targetAfterPause = getTarget();
     } catch (Exception e) {
       throw new PauserException(
           "Failed to find the target pods to examine if the targets pods were updated during"
@@ -113,8 +138,9 @@ public PausedDuration pause(int pauseDuration, @Nullable Long maxPauseWaitTime)
     return new PausedDuration(startTime, endTime);
   }
 
-  private void unpauseWithRetry(
-      RequestCoordinator coordinator, int maxRetryCount, TargetSnapshot target) {
+  @VisibleForTesting
+  void unpauseWithRetry(RequestCoordinator coordinator, int maxRetryCount, TargetSnapshot target)
+      throws PauserException {
     int retryCounter = 0;
 
     while (true) {
@@ -123,17 +149,34 @@ private void unpauseWithRetry(
         return;
       } catch (Exception e) {
         if (++retryCounter >= maxRetryCount) {
-          logger.warn(
+          // Users who directly utilize this library, bypassing our CLI, are responsible for proper
+          // exception handling. However, this scenario represents a critical issue. Consequently,
+          // we output the error message here regardless of whether the calling code handles the
+          // exception.
+          logger.error(
               "Failed to unpause Scalar product. They are still in paused. You must restart related"
                   + " pods by using the `kubectl rollout restart deployment {}`"
-                  + " command to unpase all pods.",
+                  + " command to unpause all pods.",
               target.getDeployment().getMetadata().getName());
-          return;
+          // In our CLI, we catch this exception and output the message as an error on the CLI side.
+          throw new PauserException(
+              String.format(
+                  "Failed to unpause Scalar product. They are still in paused. You must restart"
+                      + " related pods by using the `kubectl rollout restart deployment %s` command"
+                      + " to unpause all pods.",
+                  target.getDeployment().getMetadata().getName()),
+              e);
         }
       }
     }
   }
 
+  @VisibleForTesting
+  TargetSnapshot getTarget() throws PauserException {
+    return targetSelector.select();
+  }
+
+  @VisibleForTesting
   RequestCoordinator getRequestCoordinator(TargetSnapshot target) {
     return new RequestCoordinator(
         target.getPods().stream()