[ML] Unused state remover should also account for jobs in index (#37119)

The unused state remover was never adjusted to account for jobs stored
in the config index. The result was that when triggered it removed
state for all jobs stored in the config index.
This commit fixes the issue.

Closes #37109

Author: dimitris-athanasiou

x-pack/plugin/ml/qa/native-multi-node-tests/src/test/java/org/elasticsearch/xpack/ml/integration/DeleteExpiredDataIT.java

@@ -43,7 +43,6 @@
 import static org.hamcrest.Matchers.greaterThan;
 import static org.hamcrest.Matchers.greaterThanOrEqualTo;
 import static org.hamcrest.Matchers.is;
-import static org.hamcrest.Matchers.lessThan;
 import static org.hamcrest.Matchers.lessThanOrEqualTo;
 
 public class DeleteExpiredDataIT extends MlNativeAutodetectIntegTestCase {
@@ -244,7 +243,10 @@ public void testDeleteExpiredData() throws Exception {
                 .setFetchSource(false)
                 .setSize(10000)
                 .get();
-        assertThat(stateDocsResponse.getHits().getTotalHits().value, lessThan(10000L));
+
+        // Assert at least one state doc for each job
+        assertThat(stateDocsResponse.getHits().getTotalHits().value, greaterThanOrEqualTo(5L));
+
         for (SearchHit hit : stateDocsResponse.getHits().getHits()) {
             assertThat(hit.getId().startsWith("non_existing_job"), is(false));
         }

x-pack/plugin/ml/src/main/java/org/elasticsearch/xpack/ml/job/retention/UnusedStateRemover.java

@@ -14,15 +14,18 @@
 import org.elasticsearch.client.Client;
 import org.elasticsearch.cluster.service.ClusterService;
 import org.elasticsearch.xpack.core.ml.MlMetadata;
+import org.elasticsearch.xpack.core.ml.job.config.Job;
 import org.elasticsearch.xpack.core.ml.job.persistence.AnomalyDetectorsIndex;
 import org.elasticsearch.xpack.core.ml.job.persistence.ElasticsearchMappings;
 import org.elasticsearch.xpack.core.ml.job.process.autodetect.state.CategorizerState;
 import org.elasticsearch.xpack.core.ml.job.process.autodetect.state.ModelState;
 import org.elasticsearch.xpack.core.ml.job.process.autodetect.state.Quantiles;
+import org.elasticsearch.xpack.ml.job.persistence.BatchedJobsIterator;
 import org.elasticsearch.xpack.ml.job.persistence.BatchedStateDocIdsIterator;
 
 import java.util.Arrays;
 import java.util.Deque;
+import java.util.HashSet;
 import java.util.List;
 import java.util.Objects;
 import java.util.Set;
@@ -81,7 +84,18 @@ private BulkRequestBuilder findUnusedStateDocs() {
     }
 
     private Set<String> getJobIds() {
-        return MlMetadata.getMlMetadata(clusterService.state()).getJobs().keySet();
+        Set<String> jobIds = new HashSet<>();
+
+        // TODO Once at 8.0, we can stop searching for jobs in cluster state
+        // and remove cluster service as a member all together.
+        jobIds.addAll(MlMetadata.getMlMetadata(clusterService.state()).getJobs().keySet());
+
+        BatchedJobsIterator jobsIterator = new BatchedJobsIterator(client, AnomalyDetectorsIndex.configIndexName());
+        while (jobsIterator.hasNext()) {
+            Deque<Job.Builder> jobs = jobsIterator.next();
+            jobs.stream().map(Job.Builder::getId).forEach(jobIds::add);
+        }
+        return jobIds;
     }
 
     private void executeDeleteUnusedStateDocs(BulkRequestBuilder deleteUnusedStateRequestBuilder, ActionListener<Boolean> listener) {