jakthom · jakthom · Oct 23, 2024 · Oct 23, 2024 · Oct 23, 2024 · Oct 23, 2024
diff --git a/.VERSION b/.VERSION
@@ -1 +1 @@
-0.5.3
+0.6.0
diff --git a/README.md b/README.md
@@ -190,6 +190,19 @@ globalLabels:
   - env: dev
 ```
 
+### Packages
+
+Hercules includes a yml-based package loader which means extensions, macros, sources, and metrics can be logically grouped and distributed.
+
+Starter packages can be found in the [hercules-packages](/hercules-packages/) directory.
+
+**Example package registration**
+
+```
+packages:
+  - location: hercules-packages/snowflake-performance.yml
+```
+
 
 ### Embedded Analytics
 

diff --git a/cmd/hercules/app.go b/cmd/hercules/app.go
@@ -12,6 +12,7 @@ import (
 
 	"github.com/dbecorp/hercules/pkg/config"
 	"github.com/dbecorp/hercules/pkg/flock"
+	herculespackage "github.com/dbecorp/hercules/pkg/herculesPackage"
 	metrics "github.com/dbecorp/hercules/pkg/metrics"
 	"github.com/dbecorp/hercules/pkg/middleware"
 	"github.com/prometheus/client_golang/prometheus"
@@ -26,6 +27,7 @@ var VERSION string
 type Hercules struct {
 	config         config.Config
 	db             *sql.DB
+	packages       []herculespackage.Package
 	conn           *sql.Conn
 	metricRegistry *metrics.MetricRegistry
 }
@@ -48,21 +50,49 @@ func (d *Hercules) initializeFlock() {
 	d.db, d.conn = flock.InitializeDB(d.config)
 }
 
-func (d *Hercules) initializeSources() {
-	for _, source := range d.config.Sources {
-		source.InitializeWithConnection(d.conn)
+func (d *Hercules) loadPackages() {
+	pkgs := []herculespackage.Package{}
+	for _, pkgConfig := range d.config.Packages {
+		pkg, err := pkgConfig.GetPackage()
+		if err != nil {
+			log.Error().Err(err).Msg("could not get package")
+		}
+		pkgs = append(pkgs, pkg)
+	}
+	// Represent core configuration via a package
+	pkgs = append(pkgs, herculespackage.Package{
+		Name:       "core",
+		Version:    "1.0.0",
+		Extensions: d.config.Extensions,
+		Macros:     d.config.Macros,
+		Sources:    d.config.Sources,
+		Metrics:    d.config.Metrics,
+	})
+	d.packages = pkgs
+
+}
+
+func (d *Hercules) initializePackages() {
+	for _, p := range d.packages {
+		p.InitializeWithConnection(d.conn)
 	}
 }
 
 func (d *Hercules) initializeRegistry() {
-	d.metricRegistry = metrics.NewMetricRegistry(d.config.Metrics, d.config.InstanceLabels())
+	// Merge metric definitions from all packages
+	metricDefinitions := metrics.MetricDefinitions{}
+	for _, pkg := range d.packages {
+		metricDefinitions.Merge(pkg.Metrics)
+	}
+	d.metricRegistry = metrics.NewMetricRegistry(metricDefinitions, d.config.InstanceLabels())
 }
 
 func (d *Hercules) Initialize() {
 	log.Debug().Msg("initializing Hercules")
 	d.configure()
 	d.initializeFlock()
-	d.initializeSources()
+	d.loadPackages()
+	d.initializePackages()
 	d.initializeRegistry()
 	log.Debug().Interface("config", d.config).Msg("running with config")
 }

diff --git a/hercules-packages/example-nyc-taxi.yml b/hercules-packages/example-nyc-taxi.yml
@@ -0,0 +1,18 @@
+name: nyc-taxi
+version: 1.0.0
+
+sources:
+  - name: nyc_yellow_taxi_june_2024
+    type: parquet
+    source: https://d37ci6vzurychx.cloudfront.net/trip-data/yellow_tripdata_2024-07.parquet
+    materialize: true
+    refreshIntervalSeconds: 100
+
+metrics:
+  gauge:
+    - name: nyc_pickup_location_fare_total
+      help: Total NYC fares for the month of August by pickup location
+      enabled: True
+      sql: select struct_pack(pickupLocation := PULocationID::text), sum(fare_amount) as val from nyc_yellow_taxi_june_2024 group by 1
+      labels:
+        - pickupLocation
diff --git a/hercules-packages/example-tpch.yml b/hercules-packages/example-tpch.yml
@@ -0,0 +1,2 @@
+name: tpch
+version: 1.0.0
diff --git a/hercules-packages/snowflake-performance.yml b/hercules-packages/snowflake-performance.yml
@@ -0,0 +1,120 @@
+name: snowflake-performance
+version: 1.0.0
+
+extensions:
+  core:
+    - name: inet
+
+  community:
+    - name: chsql # Clickhouse macros and functions
+
+macros:
+  - sql: create or replace macro one() AS (SELECT 1);
+
+sources:
+  - name: snowflake_query_history
+    type: parquet
+    source: assets/snowflake_query_history.parquet
+    materialize: true
+    refreshIntervalSeconds: 5
+
+metrics:
+  gauge:
+    - name: query_status_count
+      help: Queries executed and their associated status, by user and warehouse
+      enabled: true
+      sql: from snowflake_query_history select struct_pack(user :=  user_name, warehouse := warehouse_name, status := lower(execution_status)) as labels, count(*) as value group by 1;
+      labels:
+        - user
+        - warehouse
+        - status
+
+    - name: queries_this_week_total
+      help: Queries this week total, by user and warehouse
+      enabled: true
+      sql: select struct_pack(user :=  user_name, warehouse := warehouse_name) as labels, count(*) as value from snowflake_query_history group by 1;
+      labels:
+        - user
+        - warehouse
+
+    - name: avg_query_duration_seconds
+      help: The average query duration for a particular user, using a particular warehouse
+      enabled: true
+      sql: select struct_pack(user :=  user_name, warehouse := warehouse_name) as labels, avg(TOTAL_ELAPSED_TIME) as value from snowflake_query_history group by 1;
+      labels:
+        - user
+        - warehouse
+
+    - name: table_operations_count
+      help: The number of operations on each table over the last week
+      enabled: true
+      sql: select struct_pack(user := user_name, query_type := query_type) as labels, count(*) as value from snowflake_query_history group by 1;
+      labels:
+        - user
+        - query_type
+
+    - name: avg_virtual_warehouse_spill_to_local_storage_bytes
+      help: The average bytes spilled to disk for queries on a specific warehouse
+      enabled: true
+      sql: select struct_pack(user :=  user_name, warehouse := warehouse_name) as labels, avg(BYTES_SPILLED_TO_LOCAL_STORAGE) as value from snowflake_query_history group by 1;
+      labels:
+        - user
+        - warehouse
+
+    - name: avg_virtual_warehouse_spill_to_remote_storage_bytes
+      help: The average bytes spilled to remote disk for queries on a specific warehouse
+      enabled: true
+      sql: select struct_pack(user :=  user_name, warehouse := warehouse_name) as labels, avg(BYTES_SPILLED_TO_REMOTE_STORAGE) as value from snowflake_query_history group by 1;
+      labels:
+        - user
+        - warehouse
+
+  histogram:
+    - name: query_duration_seconds
+      help: Histogram of query duration seconds
+      sql: select struct_pack(user :=  user_name, warehouse := warehouse_name) as labels, total_elapsed_time as value from snowflake_query_history;
+      labels:
+        - user
+        - warehouse
+      buckets:
+        - 0.1
+        - 0.5
+        - 1
+        - 2
+        - 4
+        - 8
+        - 16
+        - 32
+        - 64
+        - 128
+        - 256
+        - 512
+        - 1024
+        - 2048
+        - 4096
+        - 8192
+        - 16384
+        - 32768
+
+  summary:
+    - name: virtual_warehouse_query_duration_seconds
+      help: Summary of query duration seconds
+      sql: select struct_pack(user :=  user_name, warehouse := warehouse_name) as labels, total_elapsed_time as value from snowflake_query_history;
+      labels:
+        - user
+        - warehouse
+      objectives:
+        - 0.001
+        - 0.05
+        - 0.01
+        - 0.5
+        - 0.9
+        - 0.99
+
+  counter:
+    - name: queries_executed_count
+      help: The count of queries executed by user and warehouse
+      sql: select struct_pack(user :=  user_name, warehouse := warehouse_name) as labels, 1 as value from snowflake_query_history;
+      labels:
+        - user
+        - warehous
7326
e
diff --git a/hercules-packages/snowflake-security.yml b/hercules-packages/snowflake-security.yml
@@ -0,0 +1,2 @@
+name: snowflake-security
+version: 1.0.0