Merge pull request #729 from grafana/jdb/2021-10-ingester-rf

[enterprise-metrics] Ensure production values set ingester replication factor to three

charts/enterprise-metrics/CHANGELOG.md

@@ -10,6 +10,14 @@ Entries should be ordered as follows:
 
 Entries should include a reference to the Pull Request that introduced the change.
 
+## 1.6.0
+
+* [FEATURE] Upgrade to [Grafana Enterprise Metrics v1.5.1](https://grafana.com/docs/metrics-enterprise/latest/downloads/#v151----september-21st-2021). #729
+* [CHANGE] Production values set the ingester replication factor to three to avoid data loss.
+  The resource calculations of these values already factored in this replication factor but did not apply it in the configuration.
+  If you have not reduced the compute resources in these values then this change should have no impact besides increased resilience to ingester failure.
+  If you have reduced the compute resources, consider increasing them back to the recommended values before installing this version. #729
+
 ## 1.5.6
 
 * [BUGFIX] YAML exports are no longer included as part of the Helm chart. #726

charts/enterprise-metrics/Chart.yaml

@@ -1,5 +1,5 @@
 apiVersion: v2
-version: 1.5.6
+version: 1.6.0
 appVersion: v1.5.0
 description: 'Grafana Enterprise Metrics'
 engine: gotpl

charts/enterprise-metrics/capped-large.yaml

@@ -1,6 +1,6 @@
 # These values configure the Grafana Enterprise Metrics cluster to
 # handle production ingestion of ~10M active series using the blocks
-# storage engine.
+# storage engine scraped with a 15s interval.
 # Query requirements can vary dramatically depending on query rate and query
 # ranges. The values here satisfy a "usual" query load as seen from our
 # production clusters at this scale.
@@ -11,6 +11,14 @@
 # for more information.
 # Minio is no longer enabled and you are encouraged to use your cloud providers
 # object storage service for production deployments.
+# Ingesters are configured with a replication factor of 3 to ensure that a single ingester failur does not
+# result in data loss.
+config:
+  ingester:
+    lifecycler:
+      ring:
+        replication_factor: 3
+
 admin_api:
   replicas: 3
   resources:

charts/enterprise-metrics/capped-small.yaml

@@ -1,6 +1,6 @@
 # These values configure the Grafana Enterprise Metrics cluster to
 # handle production ingestion of ~1M active series using the blocks
-# storage engine.
+# storage engine scraped with a 15s interval.
 # Query requirements can vary dramatically depending on query rate and query
 # ranges. The values here satisfy a "usual" query load as seen from our
 # production clusters at this scale.
@@ -11,6 +11,14 @@
 # for more information.
 # Minio is no longer enabled and you are encouraged to use your cloud providers
 # object storage service for production deployments.
+# Ingesters are configured with a replication factor of 3 to ensure that a single ingester failur does not
+# result in data loss.
+config:
+  ingester:
+    lifecycler:
+      ring:
+        replication_factor: 3
+
 admin_api:
   replicas: 3
   resources:

charts/enterprise-metrics/large.yaml

@@ -1,6 +1,6 @@
 # These values configure the Grafana Enterprise Metrics cluster to
 # handle production ingestion of ~10M active series using the blocks
-# storage engine.
+# storage engine scraped with a 15s interval.
 # Query requirements can vary dramatically depending on query rate and query
 # ranges. The values here satisfy a "usual" query load as seen from our
 # production clusters at this scale.
@@ -11,6 +11,14 @@
 # for more information.
 # Minio is no longer enabled and you are encouraged to use your cloud providers
 # object storage service for production deployments.
+# Ingesters are configured with a replication factor of 3 to ensure that a single ingester failur does not
+# result in data loss.
+config:
+  ingester:
+    lifecycler:
+      ring:
+        replication_factor: 3
+
 admin_api:
   replicas: 3
   resources:

charts/enterprise-metrics/small.yaml

@@ -1,6 +1,6 @@
 # These values configure the Grafana Enterprise Metrics cluster to
 # handle production ingestion of ~1M active series using the blocks
-# storage engine.
+# storage engine scraped with a 15s interval.
 # Query requirements can vary dramatically depending on query rate and query
 # ranges. The values here satisfy a "usual" query load as seen from our
 # production clusters at this scale.
@@ -11,6 +11,14 @@
 # for more information.
 # Minio is no longer enabled and you are encouraged to use your cloud providers
 # object storage service for production deployments.
+# Ingesters are configured with a replication factor of 3 to ensure that a single ingester failur does not
+# result in data loss.
+config:
+  ingester:
+    lifecycler:
+      ring:
+        replication_factor: 3
+
 admin_api:
   replicas: 3
   resources: