Use an error to indicate stopping investigations.

The error can convey more information than a simple bool.

If it's nil = all good, otherwise the error can be returned why the
investigation was stopped.

Author: bergmannf

cadctl/cmd/investigate/investigate.go

@@ -120,8 +120,8 @@ func run(cmd *cobra.Command, _ []string) error {
 		}
 		return err
 	}
-	if result.StopInvestigations {
-		return nil
+	if result.StopInvestigations != nil {
+		return result.StopInvestigations
 	}
 
 	ccamInvestigation := ccam.Investigation{}

pkg/investigations/investigation/investigation.go

@@ -24,8 +24,8 @@ type InvestigationResult struct {
 	ServiceLogSent     InvestigationStep
 
 	// If multiple investigations might be run this can indicate a fatal error that makes running additional investigations useless.
-	// Uses a false-default so it must be set expclicitly by an investigation.
-	StopInvestigations bool
+	// If nil, investigations should continue. If not nil, should contain a meaningful error message explaining why investigations must stop.
+	StopInvestigations error
 }
 
 func NewResourceBuilder(pdClient pagerduty.Client, clusterId string, name string, logLevel string, pipelineName string) (ResourceBuilder, error) {

pkg/investigations/precheck/precheck.go

@@ -1,6 +1,8 @@
 package precheck
 
 import (
+	"errors"
+
 	cmv1 "github.com/openshift-online/ocm-sdk-go/clustersmgmt/v1"
 	investigation "github.com/openshift/configuration-anomaly-detection/pkg/investigations/investigation"
 	"github.com/openshift/configuration-anomaly-detection/pkg/logging"
@@ -23,13 +25,13 @@ func (c *ClusterStatePrecheck) Run(rb investigation.ResourceBuilder) (investigat
 	ocmClient := r.OcmClient
 	if cluster.State() == cmv1.ClusterStateUninstalling {
 		logging.Info("Cluster is uninstalling and requires no investigation. Silencing alert.")
-		result.StopInvestigations = true
+		result.StopInvestigations = errors.New("cluster is already uninstalling")
 		return result, pdClient.SilenceIncidentWithNote("CAD: Cluster is already uninstalling, silencing alert.")
 	}
 
 	if cluster.AWS() == nil {
 		logging.Info("Cloud provider unsupported, forwarding to primary.")
-		result.StopInvestigations = true
+		result.StopInvestigations = errors.New("unsupported cloud provider (non-AWS)")
 		return result, pdClient.EscalateIncidentWithNote("CAD could not run an automated investigation on this cluster: unsupported cloud provider.")
 	}
 
@@ -39,7 +41,7 @@ func (c *ClusterStatePrecheck) Run(rb investigation.ResourceBuilder) (investigat
 	}
 	if isAccessProtected {
 		logging.Info("Cluster is access protected. Escalating alert.")
-		result.StopInvestigations = true
+		result.StopInvestigations = errors.New("cluster is access protected")
 		return result, pdClient.EscalateIncidentWithNote("CAD is unable to run against access protected clusters. Please investigate.")
 	}
 	return result, nil

pkg/investigations/precheck/precheck_test.go

@@ -1,6 +1,7 @@
 package precheck
 
 import (
+	"errors"
 	"reflect"
 	"testing"
 
@@ -35,7 +36,7 @@ func TestInvestigation_Run(t *testing.T) {
 					PdClient:  pdClient,
 				},
 			}},
-			investigation.InvestigationResult{StopInvestigations: true},
+			investigation.InvestigationResult{StopInvestigations: errors.New("unsupported cloud provider (non-AWS)")},
 			false,
 		},
 	}