Don't schedule SLM jobs when services have been stopped (elastic#48658)

This adds a guard for the SLM lifecycle and retention service that
prevents new jobs from being scheduled once the service has been
stopped. Previous if the node were shut down the service would be
stopped, but a cluster state or local master election would cause a job
to attempt to be scheduled. This could lead to an uncaught
`RejectedExecutionException`.

Resolves elastic#47749

Author: dakrone

x-pack/plugin/ilm/src/main/java/org/elasticsearch/xpack/slm/SnapshotLifecycleService.java

@@ -29,6 +29,7 @@
 import java.util.Map;
 import java.util.Optional;
 import java.util.Set;
+import java.util.concurrent.atomic.AtomicBoolean;
 import java.util.function.Supplier;
 import java.util.regex.Pattern;
 import java.util.stream.Collectors;
@@ -47,6 +48,7 @@ public class SnapshotLifecycleService implements LocalNodeMasterListener, Closea
     private final ClusterService clusterService;
     private final SnapshotLifecycleTask snapshotTask;
     private final Map<String, SchedulerEngine.Job> scheduledTasks = ConcurrentCollections.newConcurrentMap();
+    private final AtomicBoolean running = new AtomicBoolean(true);
     private volatile boolean isMaster = false;
 
     public SnapshotLifecycleService(Settings settings,
@@ -142,6 +144,10 @@ public void cleanupDeletedPolicies(final ClusterState state) {
      * the same version of a policy has already been scheduled it does not overwrite the job.
      */
     public void maybeScheduleSnapshot(final SnapshotLifecyclePolicyMetadata snapshotLifecyclePolicy) {
+        if (this.running.get() == false) {
+            return;
+        }
+
         final String jobId = getJobId(snapshotLifecyclePolicy);
         final Pattern existingJobPattern = Pattern.compile(snapshotLifecyclePolicy.getPolicy().getId() + JOB_PATTERN_SUFFIX);
 
@@ -219,6 +225,8 @@ public String executorName() {
 
     @Override
     public void close() {
-        this.scheduler.stop();
+        if (this.running.compareAndSet(true, false)) {
+            this.scheduler.stop();
+        }
     }
 }

x-pack/plugin/ilm/src/test/java/org/elasticsearch/xpack/slm/SnapshotLifecycleServiceTests.java

@@ -129,6 +129,13 @@ public void testNothingScheduledWhenNotRunning() {
             // Since the service is stopped, jobs should have been cancelled
             assertThat(sls.getScheduler().scheduledJobIds(), equalTo(Collections.emptySet()));
 
+            // No jobs should be scheduled when service is closed
+            state = createState(new SnapshotLifecycleMetadata(policies, OperationMode.RUNNING, new SnapshotLifecycleStats()));
+            sls.close();
+            sls.onMaster();
+            sls.clusterChanged(new ClusterChangedEvent("1", state, emptyState));
+            assertThat(sls.getScheduler().scheduledJobIds(), equalTo(Collections.emptySet()));
+
             threadPool.shutdownNow();
         }
     }