Update rally-* index template settings to default to Elasticsearch de…

…faults (#1312)

This commit changes the defaults for the number of shards and replicas used in the Elasticsearch datastore first introduced in #1310:
instead of hardcoding them -- when unset -- to `1p/0r`, we rely on ES defaults[1] and only attempt to override these index settings if they've
been explicitly specified in the `ini` file.

Additionally, we also remove the hardcoded `refresh_interval: 5s` setting, to allow us to take advantage of Elasticsearch's default refresh behavior ([2], [3]).

\[1\]: https://www.elastic.co/guide/en/elasticsearch/reference/current/index-modules.html#_static_index_settings
\[2\]: https://www.elastic.co/guide/en/elasticsearch/reference/current/indices-refresh.html#refresh-api-desc
\[3\]: https://www.elastic.co/guide/en/elasticsearch/reference/current/index-modules.html#dynamic-index-settings

docs/configuration.rst

@@ -69,8 +69,8 @@ The following settings are applicable only if ``datastore.type`` is set to "elas
 * ``datastore.user``: Sets the name of the Elasticsearch user for the metrics store.
 * ``datastore.password``: Sets the password of the Elasticsearch user for the metrics store.
 * ``datastore.probe.cluster_version`` (default: true): Enables automatic detection of the metric store's version.
-* ``datastore.number_of_shards`` (default: 1): The number of primary shards that the ``rally-*`` indices should have. Any updates to this setting after initial index creation will only be applied to new ``rally-*`` indices.
-* ``datastore.number_of_replicas`` (default: 0): The number of replicas each primary shard has. Defaults to 0. Any updates to this setting after initial index creation will only be applied to new ``rally-*`` indices.
+* ``datastore.number_of_shards`` (default: `Elasticsearch default value <https://www.elastic.co/guide/en/elasticsearch/reference/current/index-modules.html#_static_index_settings>`_): The number of primary shards that the ``rally-*`` indices should have. Any updates to this setting after initial index creation will only be applied to new ``rally-*`` indices.
+* ``datastore.number_of_replicas`` (default: `Elasticsearch default value <https://www.elastic.co/guide/en/elasticsearch/reference/current/index-modules.html#_static_index_settings>`_): The number of replicas each primary shard has. Defaults to . Any updates to this setting after initial index creation will only be applied to new ``rally-*`` indices.
 
 **Examples**
 

esrally/metrics.py

@@ -238,8 +238,8 @@ class IndexTemplateProvider:
 
     def __init__(self, cfg):
         self._config = cfg
-        self._number_of_shards = self._config.opts("reporting", "datastore.number_of_shards", default_value=1, mandatory=False)
-        self._number_of_replicas = self._config.opts("reporting", "datastore.number_of_replicas", default_value=0, mandatory=False)
+        self._number_of_shards = self._config.opts("reporting", "datastore.number_of_shards", default_value=None, mandatory=False)
+        self._number_of_replicas = self._config.opts("reporting", "datastore.number_of_replicas", default_value=None, mandatory=False)
         self.script_dir = self._config.opts("node", "rally.root")
 
     def metrics_template(self):
@@ -254,8 +254,15 @@ def results_template(self):
     def _read(self, template_name):
         with open("%s/resources/%s.json" % (self.script_dir, template_name), encoding="utf-8") as f:
             template = json.load(f)
-            template["settings"]["index"]["number_of_shards"] = self._number_of_shards
-            template["settings"]["index"]["number_of_replicas"] = self._number_of_replicas
+            if self._number_of_shards is not None:
+                if int(self._number_of_shards) < 1:
+                    raise exceptions.SystemSetupError(
+                        f"The setting: datastore.number_of_shards must be >= 1. Please "
+                        f"check the configuration in {self._config.config_file.location}"
+                    )
+                template["settings"]["index"]["number_of_shards"] = int(self._number_of_shards)
+            if self._number_of_replicas is not None:
+                template["settings"]["index"]["number_of_replicas"] = int(self._number_of_replicas)
             return json.dumps(template)
 
 

esrally/resources/metrics-template.json

@@ -2,8 +2,6 @@
   "index_patterns": ["rally-metrics-*"],
   "settings": {
     "index": {
-      "refresh_interval": "5s",
-      "number_of_shards": 1
     }
   },
   "mappings": {

esrally/resources/races-template.json

@@ -2,8 +2,6 @@
   "index_patterns": ["rally-races-*"],
   "settings": {
     "index": {
-      "refresh_interval": "5s",
-      "number_of_shards": 1
     }
   },
   "mappings": {

esrally/resources/results-template.json

@@ -2,8 +2,6 @@
   "index_patterns": ["rally-results-*"],
   "settings": {
     "index": {
-      "refresh_interval": "5s",
-      "number_of_shards": 1
     }
   },
   "mappings": {

tests/metrics_test.py

@@ -2340,10 +2340,10 @@ def setUp(self):
         self.cfg.add(config.Scope.application, "system", "time.start", FileRaceStoreTests.RACE_TIMESTAMP)
         self.cfg.add(config.Scope.application, "system", "race.id", FileRaceStoreTests.RACE_ID)
 
-    def test_datastore_type_elasticsearch_index_template_update(self):
+    def test_primary_and_replica_shard_count_specified_index_template_update(self):
         _datastore_type = "elasticsearch"
         _datastore_number_of_shards = random.randint(1, 100)
-        _datastore_number_of_replicas = random.randint(1, 100)
+        _datastore_number_of_replicas = random.randint(0, 100)
 
         self.cfg.add(config.Scope.applicationOverride, "reporting", "datastore.type", _datastore_type)
         self.cfg.add(config.Scope.applicationOverride, "reporting", "datastore.number_of_shards", _datastore_number_of_shards)
@@ -2361,3 +2361,90 @@ def test_datastore_type_elasticsearch_index_template_update(self):
             t = json.loads(template)
             assert t["settings"]["index"]["number_of_shards"] == _datastore_number_of_shards
             assert t["settings"]["index"]["number_of_replicas"] == _datastore_number_of_replicas
+
+    def test_primary_shard_count_specified_index_template_update(self):
+        _datastore_type = "elasticsearch"
+        _datastore_number_of_shards = random.randint(1, 100)
+
+        self.cfg.add(config.Scope.applicationOverride, "reporting", "datastore.type", _datastore_type)
+        self.cfg.add(config.Scope.applicationOverride, "reporting", "datastore.number_of_shards", _datastore_number_of_shards)
+
+        _index_template_provider = metrics.IndexTemplateProvider(self.cfg)
+
+        templates = [
+            _index_template_provider.metrics_template(),
+            _index_template_provider.races_template(),
+            _index_template_provider.results_template(),
+        ]
+
+        for template in templates:
+            t = json.loads(template)
+            assert t["settings"]["index"]["number_of_shards"] == _datastore_number_of_shards
+            with self.assertRaises(KeyError):
+                # pylint: disable=unused-variable
+                number_of_replicas = t["settings"]["index"]["number_of_replicas"]
+
+    def test_replica_shard_count_specified_index_template_update(self):
+        _datastore_type = "elasticsearch"
+        _datastore_number_of_replicas = random.randint(1, 100)
+
+        self.cfg.add(config.Scope.applicationOverride, "reporting", "datastore.type", _datastore_type)
+        self.cfg.add(config.Scope.applicationOverride, "reporting", "datastore.number_of_replicas", _datastore_number_of_replicas)
+
+        _index_template_provider = metrics.IndexTemplateProvider(self.cfg)
+
+        templates = [
+            _index_template_provider.metrics_template(),
+            _index_template_provider.races_template(),
+            _index_template_provider.results_template(),
+        ]
+
+        for template in templates:
+            t = json.loads(template)
+            assert t["settings"]["index"]["number_of_replicas"] == _datastore_number_of_replicas
+            with self.assertRaises(KeyError):
+                # pylint: disable=unused-variable
+                number_of_shards = t["settings"]["index"]["number_of_shards"]
+
+    def test_primary_shard_count_less_than_one(self):
+        _datastore_type = "elasticsearch"
+        _datastore_number_of_shards = 0
+
+        self.cfg.add(config.Scope.applicationOverride, "reporting", "datastore.type", _datastore_type)
+        self.cfg.add(config.Scope.applicationOverride, "reporting", "datastore.number_of_shards", _datastore_number_of_shards)
+        _index_template_provider = metrics.IndexTemplateProvider(self.cfg)
+
+        with self.assertRaises(exceptions.SystemSetupError) as ctx:
+            # pylint: disable=unused-variable
+            templates = [
+                _index_template_provider.metrics_template(),
+                _index_template_provider.races_template(),
+                _index_template_provider.results_template(),
+            ]
+        self.assertEqual(
+            "The setting: datastore.number_of_shards must be >= 1. Please check the configuration in "
+            f"{_index_template_provider._config.config_file.location}",
+            ctx.exception.args[0],
+        )
+
+    def test_primary_and_replica_shard_counts_passed_as_strings(self):
+        _datastore_type = "elasticsearch"
+        _datastore_number_of_shards = "200"
+        _datastore_number_of_replicas = "1"
+
+        self.cfg.add(config.Scope.applicationOverride, "reporting", "datastore.type", _datastore_type)
+        self.cfg.add(config.Scope.applicationOverride, "reporting", "datastore.number_of_shards", _datastore_number_of_shards)
+        self.cfg.add(config.Scope.applicationOverride, "reporting", "datastore.number_of_replicas", _datastore_number_of_replicas)
+
+        _index_template_provider = metrics.IndexTemplateProvider(self.cfg)
+
+        templates = [
+            _index_template_provider.metrics_template(),
+            _index_template_provider.races_template(),
+            _index_template_provider.results_template(),
+        ]
+
+        for template in templates:
+            t = json.loads(template)
+            assert t["settings"]["index"]["number_of_shards"] == 200
+            assert t["settings"]["index"]["number_of_replicas"] == 1