Work on common provider interface to make adding new providers easier

alerting/provider/custom/custom.go

@@ -4,6 +4,7 @@ import (
 	"bytes"
 	"fmt"
 	"github.com/TwinProduction/gatus/client"
+	"github.com/TwinProduction/gatus/core"
 	"io/ioutil"
 	"net/http"
 	"strings"
@@ -20,6 +21,10 @@ func (provider *AlertProvider) IsValid() bool {
 	return len(provider.Url) > 0
 }
 
+func (provider *AlertProvider) ToCustomAlertProvider(service *core.Service, alert *core.Alert, result *core.Result, resolved bool) *AlertProvider {
+	return provider
+}
+
 func (provider *AlertProvider) buildRequest(serviceName, alertDescription string, resolved bool) *http.Request {
 	body := provider.Body
 	providerUrl := provider.Url

alerting/provider/pagerduty/pagerduty.go

@@ -15,7 +15,17 @@ func (provider *AlertProvider) IsValid() bool {
 }
 
 // https://developer.pagerduty.com/docs/events-api-v2/trigger-events/
-func (provider *AlertProvider) ToCustomAlertProvider(eventAction, resolveKey string, service *core.Service, message string) *custom.AlertProvider {
+func (provider *AlertProvider) ToCustomAlertProvider(service *core.Service, alert *core.Alert, result *core.Result, resolved bool) *custom.AlertProvider {
+	var message, eventAction, resolveKey string
+	if resolved {
+		message = fmt.Sprintf("RESOLVED: %s - %s", service.Name, alert.Description)
+		eventAction = "resolve"
+		resolveKey = alert.ResolveKey
+	} else {
+		message = fmt.Sprintf("TRIGGERED: %s - %s", service.Name, alert.Description)
+		eventAction = "trigger"
+		resolveKey = ""
+	}
 	return &custom.AlertProvider{
 		Url:    "https://events.pagerduty.com/v2/enqueue",
 		Method: "POST",

alerting/provider/provider.go

@@ -0,0 +1,11 @@
+package provider
+
+import (
+	"github.com/TwinProduction/gatus/alerting/provider/custom"
+	"github.com/TwinProduction/gatus/core"
+)
+
+type AlertProvider interface {
+	IsValid() bool
+	ToCustomAlertProvider(service *core.Service, alert *core.Alert, result *core.Result, resolved bool) *custom.AlertProvider
+}

alerting/provider/twilio/twilio.go

@@ -4,6 +4,7 @@ import (
 	"encoding/base64"
 	"fmt"
 	"github.com/TwinProduction/gatus/alerting/provider/custom"
+	"github.com/TwinProduction/gatus/core"
 	"net/url"
 )
 
@@ -18,7 +19,13 @@ func (provider *AlertProvider) IsValid() bool {
 	return len(provider.Token) > 0 && len(provider.SID) > 0 && len(provider.From) > 0 && len(provider.To) > 0
 }
 
-func (provider *AlertProvider) ToCustomAlertProvider(message string) *custom.AlertProvider {
+func (provider *AlertProvider) ToCustomAlertProvider(service *core.Service, alert *core.Alert, result *core.Result, resolved bool) *custom.AlertProvider {
+	var message string
+	if resolved {
+		message = fmt.Sprintf("RESOLVED: %s - %s", service.Name, alert.Description)
+	} else {
+		message = fmt.Sprintf("TRIGGERED: %s - %s", service.Name, alert.Description)
+	}
 	return &custom.AlertProvider{
 		Url:    fmt.Sprintf("https://api.twilio.com/2010-04-01/Accounts/%s/Messages.json", provider.SID),
 		Method: "POST",

client/client_test.go

@@ -0,0 +1,13 @@
+package client
+
+import "testing"
+
+func TestGetHttpClient(t *testing.T) {
+	if client != nil {
+		t.Error("client should've been nil since it hasn't been called a single time yet")
+	}
+	_ = GetHttpClient()
+	if client == nil {
+		t.Error("client shouldn't have been nil, since it has been called once")
+	}
+}

config.yaml

@@ -1,15 +1,15 @@
 metrics: true
 services:
   - name: twinnation
+    url: "https://twinnation.org/health"
     interval: 30s
-    url: https://twinnation.org/health
     conditions:
       - "[STATUS] == 200"
       - "[BODY].status == UP"
       - "[RESPONSE_TIME] < 1000"
   - name: cat-fact
-    interval: 1m
     url: "https://cat-fact.herokuapp.com/facts/random"
+    interval: 1m
     conditions:
       - "[STATUS] == 200"
       - "[BODY].deleted == false"

config/config.go

@@ -3,6 +3,7 @@ package config
 import (
 	"errors"
 	"github.com/TwinProduction/gatus/alerting"
+	"github.com/TwinProduction/gatus/alerting/provider"
 	"github.com/TwinProduction/gatus/core"
 	"gopkg.in/yaml.v2"
 	"io/ioutil"
@@ -36,7 +37,7 @@ func Get() *Config {
 }
 
 func Load(configFile string) error {
-	log.Printf("[config][Load] Attempting to load config from configFile=%s", configFile)
+	log.Printf("[config][Load] Reading configuration from configFile=%s", configFile)
 	cfg, err := readConfigurationFile(configFile)
 	if err != nil {
 		if os.IsNotExist(err) {
@@ -75,13 +76,79 @@ func parseAndValidateConfigBytes(yamlBytes []byte) (config *Config, err error) {
 	// Parse configuration file
 	err = yaml.Unmarshal(yamlBytes, &config)
 	// Check if the configuration file at least has services.
-	if config == nil || len(config.Services) == 0 {
+	if config == nil || config.Services == nil || len(config.Services) == 0 {
 		err = ErrNoServiceInConfig
 	} else {
-		// Set the default values if they aren't set
-		for _, service := range config.Services {
-			service.Validate()
-		}
+		validateAlertingConfig(config)
+		validateServicesConfig(config)
 	}
 	return
 }
+
+func validateServicesConfig(config *Config) {
+	for _, service := range config.Services {
+		if config.Debug {
+			log.Printf("[config][validateServicesConfig] Validating service '%s'", service.Name)
+		}
+		service.ValidateAndSetDefaults()
+	}
+	log.Printf("[config][validateServicesConfig] Validated %d services", len(config.Services))
+}
+
+func validateAlertingConfig(config *Config) {
+	if config.Alerting == nil {
+		log.Printf("[config][validateAlertingConfig] Alerting is not configured")
+		return
+	}
+	alertTypes := []core.AlertType{
+		core.SlackAlert,
+		core.TwilioAlert,
+		core.PagerDutyAlert,
+		core.CustomAlert,
+	}
+	var validProviders, invalidProviders []core.AlertType
+	for _, alertType := range alertTypes {
+		alertProvider := GetAlertingProviderByAlertType(config, alertType)
+		if alertProvider != nil {
+			if alertProvider.IsValid() {
+				validProviders = append(validProviders, alertType)
+			} else {
+				log.Printf("[config][validateAlertingConfig] Ignoring provider=%s because configuration is invalid", alertType)
+				invalidProviders = append(invalidProviders, alertType)
+			}
+		} else {
+			invalidProviders = append(invalidProviders, alertType)
+		}
+	}
+	log.Printf("[config][validateAlertingConfig] configuredProviders=%s; ignoredProviders=%s", validProviders, invalidProviders)
+}
+
+func GetAlertingProviderByAlertType(config *Config, alertType core.AlertType) provider.AlertProvider {
+	switch alertType {
+	case core.SlackAlert:
+		if config.Alerting.Slack == nil {
+			// Since we're returning an interface, we need to explicitly return nil, even if the provider itself is nil
+			return nil
+		}
+		return config.Alerting.Slack
+	case core.TwilioAlert:
+		if config.Alerting.Twilio == nil {
+			// Since we're returning an interface, we need to explicitly return nil, even if the provider itself is nil
+			return nil
+		}
+		return config.Alerting.Twilio
+	case core.PagerDutyAlert:
+		if config.Alerting.PagerDuty == nil {
+			// Since we're returning an interface, we need to explicitly return nil, even if the provider itself is nil
+			return nil
+		}
+		return config.Alerting.PagerDuty
+	case core.CustomAlert:
+		if config.Alerting.Custom == nil {
+			// Since we're returning an interface, we need to explicitly return nil, even if the provider itself is nil
+			return nil
+		}
+		return config.Alerting.Custom
+	}
+	return nil
+}

config/config_test.go

@@ -1,7 +1,6 @@
 package config
 
 import (
-	"fmt"
 	"github.com/TwinProduction/gatus/core"
 	"testing"
 	"time"
@@ -36,7 +35,6 @@ services:
 	if config.Services[1].Url != "https://api.github.com/healthz" {
 		t.Errorf("URL should have been %s", "https://api.github.com/healthz")
 	}
-	fmt.Println(config.Services[0].Interval)
 	if config.Services[0].Interval != 15*time.Second {
 		t.Errorf("Interval should have been %s", 15*time.Second)
 	}
@@ -123,6 +121,8 @@ func TestParseAndValidateConfigBytesWithAlerting(t *testing.T) {
 alerting:
   slack:
     webhook-url: "http://example.com"
+  pagerduty:
+    integration-key: "00000000000000000000000000000000"
 services:
   - name: twinnation
     url: https://twinnation.org/actuator/health
@@ -144,10 +144,19 @@ services:
 		t.Error("Metrics should've been false by default")
 	}
 	if config.Alerting == nil {
-		t.Fatal("config.AlertingConfig shouldn't have been nil")
+		t.Fatal("config.Alerting shouldn't have been nil")
+	}
+	if config.Alerting.Slack == nil || !config.Alerting.Slack.IsValid() {
+		t.Fatal("Slack alerting config should've been valid")
 	}
 	if config.Alerting.Slack.WebhookUrl != "http://example.com" {
-		t.Errorf("Slack webhook should've been %s, but was %s", "http://example.com", config.Alerting.Slack)
+		t.Errorf("Slack webhook should've been %s, but was %s", "http://example.com", config.Alerting.Slack.WebhookUrl)
+	}
+	if config.Alerting.PagerDuty == nil || !config.Alerting.PagerDuty.IsValid() {
+		t.Fatal("PagerDuty alerting config should've been valid")
+	}
+	if config.Alerting.PagerDuty.IntegrationKey != "00000000000000000000000000000000" {
+		t.Errorf("PagerDuty integration key should've been %s, but was %s", "00000000000000000000000000000000", config.Alerting.PagerDuty.IntegrationKey)
 	}
 	if len(config.Services) != 1 {
 		t.Error("There should've been 1 service")
@@ -180,3 +189,31 @@ services:
 		t.Errorf("The type of the alert should've been %s, but it was %s", "Healthcheck failed 7 times in a row", config.Services[0].Alerts[0].Description)
 	}
 }
+
+func TestParseAndValidateConfigBytesWithInvalidPagerDutyAlertingConfig(t *testing.T) {
+	config, err := parseAndValidateConfigBytes([]byte(`
+alerting:
+  pagerduty:
+    integration-key: "INVALID_KEY"
+services:
+  - name: twinnation
+    url: https://twinnation.org/actuator/health
+    conditions:
+      - "[STATUS] == 200"
+`))
+	if err != nil {
+		t.Error("No error should've been returned")
+	}
+	if config == nil {
+		t.Fatal("Config shouldn't have been nil")
+	}
+	if config.Alerting == nil {
+		t.Fatal("config.Alerting shouldn't have been nil")
+	}
+	if config.Alerting.PagerDuty == nil {
+		t.Fatal("PagerDuty alerting config shouldn't have been nil")
+	}
+	if config.Alerting.PagerDuty.IsValid() {
+		t.Fatal("PagerDuty alerting config should've been invalid")
+	}
+}

core/service.go

@@ -50,7 +50,8 @@ type Service struct {
 	NumberOfSuccessesInARow int
 }
 
-func (service *Service) Validate() {
+// ValidateAndSetDefaults validates the service's configuration and sets the default value of fields that have one
+func (service *Service) ValidateAndSetDefaults() {
 	// Set default values
 	if service.Interval == 0 {
 		service.Interval = 1 * time.Minute
@@ -83,7 +84,8 @@ func (service *Service) Validate() {
 	}
 }
 
-func (service *Service) EvaluateConditions() *Result {
+// EvaluateHealth sends a request to the service's URL and evaluates the conditions of the service.
+func (service *Service) EvaluateHealth() *Result {
 	result := &Result{Success: true, Errors: []string{}}
 	service.getIp(result)
 	if len(result.Errors) == 0 {

core/service_test.go

@@ -4,14 +4,14 @@ import (
 	"testing"
 )
 
-func TestIntegrationEvaluateConditions(t *testing.T) {
+func TestIntegrationEvaluateHealth(t *testing.T) {
 	condition := Condition("[STATUS] == 200")
 	service := Service{
 		Name:       "TwiNNatioN",
 		Url:        "https://twinnation.org/health",
 		Conditions: []*Condition{&condition},
 	}
-	result := service.EvaluateConditions()
+	result := service.EvaluateHealth()
 	if !result.ConditionResults[0].Success {
 		t.Errorf("Condition '%s' should have been a success", condition)
 	}
@@ -20,14 +20,14 @@ func TestIntegrationEvaluateConditions(t *testing.T) {
 	}
 }
 
-func TestIntegrationEvaluateConditionsWithFailure(t *testing.T) {
+func TestIntegrationEvaluateHealthWithFailure(t *testing.T) {
 	condition := Condition("[STATUS] == 500")
 	service := Service{
 		Name:       "TwiNNatioN",
 		Url:        "https://twinnation.org/health",
 		Conditions: []*Condition{&condition},
 	}
-	result := service.EvaluateConditions()
+	result := service.EvaluateHealth()
 	if result.ConditionResults[0].Success {
 		t.Errorf("Condition '%s' should have been a failure", condition)
 	}

example/kubernetes/gatus.yaml

@@ -49,10 +49,10 @@ spec:
               protocol: TCP
           resources:
             limits:
-              cpu: 50m
+              cpu: 200m
               memory: 50M
             requests:
-              cpu: 20m
+              cpu: 50m
               memory: 20M
           volumeMounts:
             - mountPath: /config

watchdog/alerting.go

@@ -2,8 +2,6 @@ package watchdog
 
 import (
 	"encoding/json"
-	"fmt"
-	"github.com/TwinProduction/gatus/alerting/provider/custom"
 	"github.com/TwinProduction/gatus/config"
 	"github.com/TwinProduction/gatus/core"
 	"log"
@@ -36,42 +34,16 @@ func handleAlertsToTrigger(service *core.Service, result *core.Result, cfg *conf
 			}
 			continue
 		}
-		var alertProvider *custom.AlertProvider
-		if alert.Type == core.SlackAlert {
-			if cfg.Alerting.Slack != nil && cfg.Alerting.Slack.IsValid() {
-				log.Printf("[watchdog][handleAlertsToTrigger] Sending Slack alert because alert with description='%s' has been triggered", alert.Description)
-				alertProvider = cfg.Alerting.Slack.ToCustomAlertProvider(service, alert, result, false)
-			} else {
-				log.Printf("[watchdog][handleAlertsToTrigger] Not sending Slack alert despite being triggered, because there is no Slack webhook configured")
-			}
-		} else if alert.Type == core.PagerDutyAlert {
-			if cfg.Alerting.PagerDuty != nil && cfg.Alerting.PagerDuty.IsValid() {
-				log.Printf("[watchdog][handleAlertsToTrigger] Sending PagerDuty alert because alert with description='%s' has been triggered", alert.Description)
-				alertProvider = cfg.Alerting.PagerDuty.ToCustomAlertProvider("trigger", "", service, fmt.Sprintf("TRIGGERED: %s - %s", service.Name, alert.Description))
-			} else {
-				log.Printf("[watchdog][handleAlertsToTrigger] Not sending PagerDuty alert despite being triggered, because PagerDuty isn't configured properly")
-			}
-		} else if alert.Type == core.TwilioAlert {
-			if cfg.Alerting.Twilio != nil && cfg.Alerting.Twilio.IsValid() {
-				log.Printf("[watchdog][handleAlertsToTrigger] Sending Twilio alert because alert with description='%s' has been triggered", alert.Description)
-				alertProvider = cfg.Alerting.Twilio.ToCustomAlertProvider(fmt.Sprintf("TRIGGERED: %s - %s", service.Name, alert.Description))
-			} else {
-				log.Printf("[watchdog][handleAlertsToTrigger] Not sending Twilio alert despite being triggered, because Twilio config settings missing")
-			}
-		} else if alert.Type == core.CustomAlert {
-			if cfg.Alerting.Custom != nil && cfg.Alerting.Custom.IsValid() {
-				log.Printf("[watchdog][handleAlertsToTrigger] Sending custom alert because alert with description='%s' has been triggered", alert.Description)
-				alertProvider = cfg.Alerting.Custom
-			} else {
-				log.Printf("[watchdog][handleAlertsToTrigger] Not sending custom alert despite being triggered, because there is no custom url configured")
-			}
-		}
-		if alertProvider != nil {
+		alertProvider := config.GetAlertingProviderByAlertType(cfg, alert.Type)
+		if alertProvider != nil && alertProvider.IsValid() {
+			log.Printf("[watchdog][handleAlertsToTrigger] Sending %s alert because alert with description='%s' has been triggered", alert.Type, alert.Description)
+			customAlertProvider := alertProvider.ToCustomAlertProvider(service, alert, result, false)
 			// TODO: retry on error
 			var err error
+			// We need to extract the DedupKey from PagerDuty's response
 			if alert.Type == core.PagerDutyAlert {
 				var body []byte
-				body, err = alertProvider.Send(service.Name, alert.Description, true)
+				body, err = customAlertProvider.Send(service.Name, alert.Description, false)
 				if err == nil {
 					var response pagerDutyResponse
 					err = json.Unmarshal(body, &response)
@@ -82,13 +54,17 @@ func handleAlertsToTrigger(service *core.Service, result *core.Result, cfg *conf
 					}
 				}
 			} else {
-				_, err = alertProvider.Send(service.Name, alert.Description, false)
+				// All other alert types don't need to extract anything from the body, so we can just send the request right away
+				_, err = customAlertProvider.Send(service.Name, alert.Description, false)
 			}
 			if err != nil {
 				log.Printf("[watchdog][handleAlertsToTrigger] Ran into error sending an alert: %s", err.Error())
 			} else {
 				alert.Triggered = true
 			}
+
+		} else {
+			log.Printf("[watchdog][handleAlertsToResolve] Not sending alert of type=%s despite being triggered, because the provider wasn't configured properly", alert.Type)
 		}
 	}
 }
@@ -103,44 +79,12 @@ func handleAlertsToResolve(service *core.Service, result *core.Result, cfg *conf
 		if !alert.SendOnResolved {
 			continue
 		}
-		var alertProvider *custom.AlertProvider
-		if alert.Type == core.SlackAlert {
-			if cfg.Alerting.Slack != nil && cfg.Alerting.Slack.IsValid() {
-				log.Printf("[watchdog][handleAlertsToResolve] Sending Slack alert because alert with description='%s' has been resolved", alert.Description)
-				alertProvider = cfg.Alerting.Slack.ToCustomAlertProvider(service, alert, result, true)
-			} else {
-				log.Printf("[watchdog][handleAlertsToResolve] Not sending Slack alert despite being resolved, because there is no Slack webhook configured")
-			}
-		} else if alert.Type == core.PagerDutyAlert {
-			if cfg.Alerting.PagerDuty != nil && cfg.Alerting.PagerDuty.IsValid() {
-				log.Printf("[watchdog][handleAlertsToResolve] Sending PagerDuty alert because alert with description='%s' has been resolved", alert.Description)
-				alertProvider = cfg.Alerting.PagerDuty.ToCustomAlertProvider("resolve", alert.ResolveKey, service, fmt.Sprintf("RESOLVED: %s - %s", service.Name, alert.Description))
-			} else {
-				log.Printf("[watchdog][handleAlertsToResolve] Not sending PagerDuty alert despite being resolved, because PagerDuty isn't configured properly")
-			}
-		} else if alert.Type == core.TwilioAlert {
-			if cfg.Alerting.Twilio != nil && cfg.Alerting.Twilio.IsValid() {
-				log.Printf("[watchdog][handleAlertsToResolve] Sending Twilio alert because alert with description='%s' has been resolved", alert.Description)
-				alertProvider = cfg.Alerting.Twilio.ToCustomAlertProvider(fmt.Sprintf("RESOLVED: %s - %s", service.Name, alert.Description))
-			} else {
-				log.Printf("[watchdog][handleAlertsToResolve] Not sending Twilio alert despite being resolved, because Twilio isn't configured properly")
-			}
-		} else if alert.Type == core.CustomAlert {
-			if cfg.Alerting.Custom != nil && cfg.Alerting.Custom.IsValid() {
-				log.Printf("[watchdog][handleAlertsToResolve] Sending custom alert because alert with description='%s' has been resolved", alert.Description)
-				alertProvider = &custom.AlertProvider{
-					Url:     cfg.Alerting.Custom.Url,
-					Method:  cfg.Alerting.Custom.Method,
-					Body:    cfg.Alerting.Custom.Body,
-					Headers: cfg.Alerting.Custom.Headers,
-				}
-			} else {
-				log.Printf("[watchdog][handleAlertsToResolve] Not sending custom alert despite being resolved, because the custom provider isn't configured properly")
-			}
-		}
-		if alertProvider != nil {
+		alertProvider := config.GetAlertingProviderByAlertType(cfg, alert.Type)
+		if alertProvider != nil && alertProvider.IsValid() {
+			log.Printf("[watchdog][handleAlertsToResolve] Sending %s alert because alert with description='%s' has been resolved", alert.Type, alert.Description)
+			customAlertProvider := alertProvider.ToCustomAlertProvider(service, alert, result, true)
 			// TODO: retry on error
-			_, err := alertProvider.Send(service.Name, alert.Description, true)
+			_, err := customAlertProvider.Send(service.Name, alert.Description, true)
 			if err != nil {
 				log.Printf("[watchdog][handleAlertsToResolve] Ran into error sending an alert: %s", err.Error())
 			} else {
@@ -148,6 +92,8 @@ func handleAlertsToResolve(service *core.Service, result *core.Result, cfg *conf
 					alert.ResolveKey = ""
 				}
 			}
+		} else {
+			log.Printf("[watchdog][handleAlertsToResolve] Not sending alert of type=%s despite being resolved, because the provider wasn't configured properly", alert.Type)
 		}
 	}
 	service.NumberOfFailuresInARow = 0

watchdog/watchdog.go

@@ -50,7 +50,7 @@ func monitor(service *core.Service) {
 		if cfg.Debug {
 			log.Printf("[watchdog][monitor] Monitoring serviceName=%s", service.Name)
 		}
-		result := service.EvaluateConditions()
+		result := service.EvaluateHealth()
 		metric.PublishMetricsForService(service, result)
 		serviceResultsMutex.Lock()
 		serviceResults[service.Name] = append(serviceResults[service.Name], result)