Convert license struct to MapStr

CHANGELOG.next.asciidoc

@@ -248,6 +248,7 @@ https://github.com/elastic/beats/compare/v7.0.0-alpha2...master[Check the HEAD d
 - Fix ARN parsing function to work for ELB ARNs. {pull}14316[14316]
 - Update azure configuration example. {issue}14224[14224]
 - Fix cloudwatch metricset with names and dimensions in config. {issue}14376[14376] {pull}14391[14391]
+- Fix marshaling of ms-since-epoch values in `elasticsearch/cluster_stats` metricset. {pull}14378[14378]
 
 *Packetbeat*
 

metricbeat/module/elasticsearch/cluster_stats/data_xpack.go

@@ -33,11 +33,6 @@ import (
 	"github.com/elastic/beats/metricbeat/module/elasticsearch"
 )
 
-type clusterStatsLicense struct {
-	*elasticsearch.License
-	ClusterNeedsTLS bool `json:"cluster_needs_tls"`
-}
-
 func clusterNeedsTLSEnabled(license *elasticsearch.License, stackStats common.MapStr) (bool, error) {
 	// TLS does not need to be enabled if license type is something other than trial
 	if !license.IsOneOf("trial") {
@@ -205,7 +200,8 @@ func eventMappingXPack(r mb.ReporterV2, m *MetricSet, info elasticsearch.Info, c
 		return errors.Wrap(err, "failed to determine if cluster needs TLS enabled")
 	}
 
-	l := clusterStatsLicense{license, clusterNeedsTLS}
+	l := license.ToMapStr()
+	l["cluster_needs_tls"] = clusterNeedsTLS
 
 	isAPMFound, err := apmIndicesExist(clusterState)
 	if err != nil {

metricbeat/module/elasticsearch/elasticsearch.go

@@ -480,6 +480,26 @@ func (l *License) IsOneOf(candidateLicenses ...string) bool {
 	return false
 }
 
+// ToMapStr converts the license to a common.MapStr. This is necessary
+// for proper marshaling of the data before it's sent over the wire. In
+// particular it ensures that ms-since-epoch values are marshaled as longs
+// and not floats in scientific notation as Elasticsearch does not like that.
+func (l *License) ToMapStr() common.MapStr {
+	return common.MapStr{
+		"status":                l.Status,
+		"id":                    l.ID,
+		"type":                  l.Type,
+		"issue_date":            l.IssueDate,
+		"issue_date_in_millis":  l.IssueDateInMillis,
+		"expiry_date":           l.ExpiryDate,
+		"expiry_date_in_millis": l.ExpiryDateInMillis,
+		"max_nodes":             l.MaxNodes,
+		"issued_to":             l.IssuedTo,
+		"issuer":                l.Issuer,
+		"start_date_in_millis":  l.StartDateInMillis,
+	}
+}
+
 func getSettingGroup(allSettings common.MapStr, groupKey string) (common.MapStr, error) {
 	hasSettingGroup, err := allSettings.HasKey(groupKey)
 	if err != nil {

metricbeat/module/elasticsearch/test_elasticsearch.py

@@ -109,6 +109,46 @@ def test_xpack(self):
         proc.check_kill_and_wait()
         self.assert_no_logged_warnings()
 
+    @unittest.skipUnless(metricbeat.INTEGRATION_TESTS, "integration test")
+    def test_xpack_cluster_stats(self):
+        """
+        elasticsearch-xpack module test for type:cluster_stats
+        """
+        self.render_config_template(modules=[{
+            "name": "elasticsearch",
+            "metricsets": [
+                "ccr",
+                "cluster_stats",
+                "enrich",
+                "index",
+                "index_recovery",
+                "index_summary",
+                "ml_job",
+                "node_stats",
+                "shard"
+            ],
+            "hosts": self.get_hosts(),
+            "period": "1s",
+            "extras": {
+                "xpack.enabled": "true"
+            }
+        }])
+        proc = self.start_beat()
+        self.wait_log_contains('"type": "cluster_stats"')
+
+        # self.wait_until(lambda: self.output_has_message('"type":"cluster_stats"'))
+        proc.check_kill_and_wait()
+        self.assert_no_logged_warnings()
+
+        docs = self.read_output_json()
+        for doc in docs:
+            t = doc["type"]
+            if t != "cluster_stats":
+                continue
+            license = doc["license"]
+            issue_date = license["issue_date_in_millis"]
+            self.assertIsNot(type(issue_date), float)
+
     def create_ml_job(self):
         # Check if an ml job already exists
         response = self.ml_es.get_jobs()