Track cluster identifier (cluster ID) as part of system information (#…

…616)

* Track cluster identifier (cluster ID) as part of system information

This is intended to help identify servers that belong together, to allow
the pganalyze app to group them in an overview, offer data in combined
views, or allow Index Advisor to avoid considering indexes unused if
they are used on a replica.

To start with this identifies the cluster ID for given providers:
- AWS: DBClusterIdentifier (user-assigned cluster name)
- Azure: Source Server ID (replication parent server ID + resource group)
- Crunchy Bridge: Cluster Parent ID (replication parent server ID)
- Self hosted: pg_controldata system identifier (unique ID shared by all physical replicas)

There are providers (e.g. GCP) and edge cases (e.g. when helper doesn't
run for self hosted) where this doesn't work. Handling those cases, and
allowing manual overrides of the cluster ID is left for a later follow-up.

input/system/azure/system.go

@@ -2,6 +2,7 @@ package azure
 
 import (
 	"context"
+	"fmt"
 	"strings"
 
 	"github.com/Azure/azure-sdk-for-go/sdk/azcore/arm"
@@ -61,6 +62,14 @@ func GetSystemState(ctx context.Context, server *state.Server, logger *util.Logg
 
 				if config.AzureDbServerName == rID.Name {
 					customWindowEnabled := util.StringCustomTypePtrToString(v.Properties.MaintenanceWindow.CustomWindow) == "Enabled"
+					if v.Properties.SourceServerResourceID != nil {
+						sourceID, err := arm.ParseResourceID(*v.Properties.SourceServerResourceID)
+						if err == nil {
+							system.Info.ClusterID = fmt.Sprintf("%s/%s", sourceID.ResourceGroupName, sourceID.Name)
+						}
+					} else {
+						system.Info.ClusterID = fmt.Sprintf("%s/%s", rID.ResourceGroupName, rID.Name)
+					}
 					system.Info.Azure = &state.SystemInfoAzure{
 						Location:                util.StringPtrToString(v.Location),
 						CreatedAt:               util.TimePtrToTime(v.SystemData.CreatedAt),
@@ -118,6 +127,7 @@ func GetSystemState(ctx context.Context, server *state.Server, logger *util.Logg
 
 					if config.AzureDbServerName == rID.Name {
 						customWindowEnabled := util.StringCustomTypePtrToString(v.Properties.MaintenanceWindow.CustomWindow) == "Enabled"
+						system.Info.ClusterID = fmt.Sprintf("%s/%s", rID.ResourceGroupName, rID.Name)
 						system.Info.Azure = &state.SystemInfoAzure{
 							Location:                 util.StringPtrToString(v.Location),
 							CreatedAt:                util.TimePtrToTime(v.SystemData.CreatedAt),

input/system/crunchy_bridge/client.go

@@ -7,6 +7,7 @@ import (
 	"io"
 	"net/http"
 
+	"github.com/guregu/null"
 	"github.com/pganalyze/collector/util"
 )
 
@@ -21,14 +22,15 @@ type Client struct {
 }
 
 type ClusterInfo struct {
-	CPU        int32   `json:"cpu"`
-	CreatedAt  string  `json:"created_at"`
-	Memory     float32 `json:"memory"`
-	Name       string  `json:"name"`
-	PlanID     string  `json:"plan_id"`
-	ProviderID string  `json:"provider_id"`
-	RegionID   string  `json:"region_id"`
-	Storage    int32   `json:"storage"`
+	CPU        int32       `json:"cpu"`
+	CreatedAt  string      `json:"created_at"`
+	Memory     float32     `json:"memory"`
+	Name       string      `json:"name"`
+	ParentID   null.String `json:"parent_id"`
+	PlanID     string      `json:"plan_id"`
+	ProviderID string      `json:"provider_id"`
+	RegionID   string      `json:"region_id"`
+	Storage    int32       `json:"storage"`
 }
 
 type MetricViews struct {

input/system/crunchy_bridge/system.go

@@ -32,6 +32,11 @@ func GetSystemState(ctx context.Context, server *state.Server, logger *util.Logg
 		logger.PrintError("CrunchyBridge/System: Encountered error when getting cluster info %v\n", err)
 		return
 	}
+	if clusterInfo.ParentID.Valid {
+		system.Info.ClusterID = clusterInfo.ParentID.String
+	} else {
+		system.Info.ClusterID = client.ClusterID
+	}
 	system.Info.CrunchyBridge = &state.SystemInfoCrunchyBridge{
 		ClusterName: clusterInfo.Name,
 		PlanID:      clusterInfo.PlanID,

input/system/rds/system.go

@@ -38,6 +38,8 @@ func GetSystemState(server *state.Server, logger *util.Logger) (system state.Sys
 		return
 	}
 
+	system.Info.ClusterID = util.StringPtrToString(instance.DBClusterIdentifier)
+
 	isAurora := util.StringPtrToString(instance.Engine) == "aurora-postgresql"
 
 	system.Info.AmazonRds = &state.SystemInfoAmazonRds{

input/system/selfhosted/system.go

@@ -49,6 +49,7 @@ func GetSystemState(server *state.Server, logger *util.Logger) (system state.Sys
 
 		system.XlogUsedBytes = status.XlogUsedBytes
 		system.Info.SelfHosted.DatabaseSystemIdentifier = status.SystemIdentifier
+		system.Info.ClusterID = status.SystemIdentifier
 	}
 
 	hostInfo, err := host.Info()

output/transform/system.go

@@ -154,6 +154,7 @@ func transformSystem(systemState state.SystemState, diffState state.DiffState) *
 
 	system.SystemId = systemState.Info.SystemID
 	system.SystemScope = systemState.Info.SystemScope
+	system.ClusterId = systemState.Info.ClusterID
 	system.XlogUsedBytes = systemState.XlogUsedBytes
 
 	system.SchedulerStatistic = &snapshot.SchedulerStatistic{

protobuf/shared.proto

@@ -96,6 +96,7 @@ message System {
   SystemInformation system_information = 1;
   string system_id = 2; // Unique identifier for this system
   string system_scope = 3; // Name the system ID is scoped by (optional)
+  string cluster_id = 4; // Unique identifier for the (physical replication) cluster
 
   SchedulerStatistic scheduler_statistic = 10;
   MemoryStatistic memory_statistic = 11;

state/system.go

@@ -39,6 +39,7 @@ type SystemInfo struct {
 	Type        SystemType
 	SystemScope string
 	SystemID    string
+	ClusterID   string
 
 	SelfHosted    *SystemInfoSelfHosted
 	AmazonRds     *SystemInfoAmazonRds