Prevent multiple services from being evaluated at the same time

README.md

@@ -10,11 +10,15 @@ Live example: https://status.twinnation.org/
 
 ## Usage
 
+By default, the configuration file is expected to be at `config/config.yaml`.
+
+You can specify a custom path by setting the `GATUS_CONFIG_FILE` environment variable.
+
 ```yaml
 metrics: true         # Whether to expose metrics at /metrics
 services:
   - name: twinnation  # Name of your service, can be anything
-    url: https://twinnation.org/actuator/health
+    url: https://twinnation.org/health
     interval: 15s     # Duration to wait between every status check (opt. default: 10s)
     conditions:
       - "[STATUS] == 200"

config.yaml

@@ -1,11 +1,12 @@
 metrics: true
 services:
-  - name: twinnation
-    url: https://twinnation.org/actuator/health
-    interval: 15s
+  - name: Twinnation
+    url: https://twinnation.org/health
+    interval: 30s
     conditions:
       - "[STATUS] == 200"
-  - name: github
+  - name: GitHub API
     url: https://api.github.com/healthz
+    interval: 30s
     conditions:
       - "[STATUS] == 200"

core/types.go

@@ -23,7 +23,7 @@ type ServerMessage struct {
 type Result struct {
 	HttpStatus       int                `json:"status"`
 	Hostname         string             `json:"hostname"`
-	Ip               string             `json:"ip"`
+	Ip               string             `json:"-"`
 	Duration         time.Duration      `json:"duration"`
 	Errors           []string           `json:"errors"`
 	ConditionResults []*ConditionResult `json:"condition-results"`
@@ -76,7 +76,7 @@ func (service *Service) EvaluateConditions() *Result {
 		result.Success = false
 	}
 	for _, condition := range service.Conditions {
-		success := condition.Evaluate(result)
+		success := condition.evaluate(result)
 		if !success {
 			result.Success = false
 		}
@@ -88,12 +88,12 @@ func (service *Service) EvaluateConditions() *Result {
 type ConditionResult struct {
 	Condition   *Condition `json:"condition"`
 	Success     bool       `json:"success"`
-	Explanation string     `json:"explanation"`
+	Explanation string     `json:"-"`
 }
 
 type Condition string
 
-func (c *Condition) Evaluate(result *Result) bool {
+func (c *Condition) evaluate(result *Result) bool {
 	condition := string(*c)
 	if strings.Contains(condition, "==") {
 		parts := sanitizeAndResolve(strings.Split(condition, "=="), result)

core/types_test.go

@@ -7,7 +7,7 @@ import (
 func TestEvaluateWithIp(t *testing.T) {
 	condition := Condition("[IP] == 127.0.0.1")
 	result := &Result{Ip: "127.0.0.1"}
-	condition.Evaluate(result)
+	condition.evaluate(result)
 	if !result.ConditionResults[0].Success {
 		t.Errorf("Condition '%s' should have been a success", condition)
 	}
@@ -16,7 +16,7 @@ func TestEvaluateWithIp(t *testing.T) {
 func TestEvaluateWithStatus(t *testing.T) {
 	condition := Condition("[STATUS] == 201")
 	result := &Result{HttpStatus: 201}
-	condition.Evaluate(result)
+	condition.evaluate(result)
 	if !result.ConditionResults[0].Success {
 		t.Errorf("Condition '%s' should have been a success", condition)
 	}
@@ -25,7 +25,7 @@ func TestEvaluateWithStatus(t *testing.T) {
 func TestEvaluateWithFailure(t *testing.T) {
 	condition := Condition("[STATUS] == 200")
 	result := &Result{HttpStatus: 500}
-	condition.Evaluate(result)
+	condition.evaluate(result)
 	if result.ConditionResults[0].Success {
 		t.Errorf("Condition '%s' should have been a failure", condition)
 	}

example/docker-compose-grafana-prometheus/config.yaml

@@ -1,7 +1,7 @@
 metrics: true
 services:
   - name: TwiNNatioN
-    url: https://twinnation.org/actuator/health
+    url: https://twinnation.org/health
     interval: 10s
     conditions:
       - "[STATUS] == 200"

example/kubernetes/gatus.yaml

@@ -4,7 +4,7 @@ data:
     metrics: true
     services:
       - name: TwiNNatioN
-        url: https://twinnation.org/actuator/health
+        url: https://twinnation.org/health
         interval: 1m
         conditions:
           - "[STATUS] == 200"

main.go

@@ -25,10 +25,10 @@ func main() {
 }
 
 func loadConfiguration() *config.Config {
-	args := os.Args
 	var err error
-	if len(args) == 2 {
-		err = config.Load(args[1])
+	customConfigFile := os.Getenv("GATUS_CONFIG_FILE")
+	if len(customConfigFile) > 0 {
+		err = config.Load(customConfigFile)
 	} else {
 		err = config.LoadDefaultConfiguration()
 	}

static/index.html

@@ -4,11 +4,23 @@
 	<title>Status</title>
 	<link rel="stylesheet" href="https://stackpath.bootstrapcdn.com/bootstrap/4.3.1/css/bootstrap.min.css" integrity="sha384-ggOyR0iXCbMQv3Xipma34MD+dH/1fQ784/j6cY/iJTQUOhcWr7x9JvoRxT2MZw1T" crossorigin="anonymous">
 	<style>
-		td > span.badge {
+		#results div.container:first-child {
+			border-top-left-radius: 3px;
+			border-top-right-radius: 3px;
+		}
+		#results div.container:last-child {
+			border-bottom-left-radius: 3px;
+			border-bottom-right-radius: 3px;
+			border-bottom-width: 1px;
+			border-color: #dee2e6;
+			border-style: solid;
+		}
+		.status-ok {
 			display: inline-block;
-			width: 20px;
-			cursor: default;
-			margin-right: 2px;
+			width: 1%;
+			height: 20px;
+			margin-right: 4px;
+			background-color: #28a745;
 		}
 	</style>
 </head>
@@ -17,19 +29,7 @@
 		<div class="text-center mb-3">
 			<div class="display-4">Status</div>
 		</div>
-		<div class="table-responsive">
-			<table class="table">
-				<thead>
-					<tr>
-						<th scope="col">Name</th>
-						<th scope="col">Status</th>
-						<th scope="col">Hostname</th>
-						<th scope="col">Response time</th>
-					</tr>
-				</thead>
-				<tbody id="results">
-				</tbody>
-			</table>
+		<div id="results">
 		</div>
 	</div>
 
@@ -38,8 +38,9 @@
 	<script src="https://stackpath.bootstrapcdn.com/bootstrap/4.3.1/js/bootstrap.min.js" integrity="sha384-JjSmVgyd0p3pXB1rRibZUAYoIIy6OrQ6VrjIEaFf/nJGzIxFDsf4x0xIM+B07jRM" crossorigin="anonymous"></script>
 
 	<script>
-		const OK = "<span class='badge badge-success' title='__RESPONSE_TIME____CONDITIONS____ERRORS__'>&#10003;</span>";
-		const NOK = "<span class='badge badge-danger' title='__RESPONSE_TIME____CONDITIONS____ERRORS__'>X</span>";
+		//const OK = "<div class='status-ok' title='__RESPONSE_TIME____CONDITIONS____ERRORS__'></div>"
+		const OK = "<span class='badge badge-success ml-1' style='width: 5%' title='__RESPONSE_TIME____CONDITIONS____ERRORS__'>&#10003;</span>";
+		const NOK = "<span class='badge badge-danger ml-1' style='width: 5%' title='__RESPONSE_TIME____CONDITIONS____ERRORS__'>X</span>";
 
 		function generateServiceResultBox(serviceResult) {
 			let output = (serviceResult.success ? OK : NOK);
@@ -62,9 +63,9 @@
 			return output;
 		}
 		
-		function refreshTable() {
+		function refreshResults() {
 			$.getJSON("/api/v1/results", function (data) {
-				let tableBody = "";
+				let output = "";
 				for (let serviceName in data) {
 					let serviceStatusOverTime = "";
 					let hostname = data[serviceName][data[serviceName].length-1].hostname 
@@ -82,21 +83,30 @@
 							maxResponseTime = responseTime;
 						}
 					}
-					tableBody += ""
-						+ "<tr>"
-						+ "  <td>" + serviceName + "</td>"
-						+ "  <td>" + serviceStatusOverTime + "</td>"
-						+ "  <td><a href=\"//" + hostname + "\">" + hostname + "</a></td>"
-						+ "  <td>" + (minResponseTime === maxResponseTime ? minResponseTime : (minResponseTime + "-" + maxResponseTime)) + "ms</td>"
-						+ "</tr>";
+					output += ""
+						+ "<div class='container p-2 border-left border-right border-top border-black'>"
+						+ "  <div class='row mb-2'>"
+						+ "    <div class='col-8'>"
+						+ "      <span class='font-weight-bold'>" + serviceName + "</span> <span class='text-secondary font-weight-lighter'>- " + hostname + "</span>"
+						+ "    </div>"
+						+ "    <div class='col-4 text-right'>"
+						+ "      <span class='font-weight-lighter'>" + (minResponseTime === maxResponseTime ? minResponseTime : (minResponseTime + "-" + maxResponseTime)) + "ms</span>"
+						+ "    </div>"
+						+ "  </div>"
+						+ "  <div class='row'>"
+						+ "    <div class='col-12 d-flex flex-row-reverse'>"
+						+ "      " + serviceStatusOverTime
+						+ "    </div>"
+						+ "  </div>"
+						+ "</div>";
 				}
-				$("#results").html(tableBody);
+				$("#results").html(output);
 			});
 		}
 
-		refreshTable();
+		refreshResults();
 		setInterval(function() {
-			refreshTable();
+			refreshResults();
 		}, 10000);
 	</script>
 </body>

watchdog/watchdog.go

@@ -20,28 +20,32 @@ func GetServiceResults() *map[string][]*core.Result {
 
 func Monitor(cfg *config.Config) {
 	for _, service := range cfg.Services {
-		go func(service *core.Service) {
-			for {
-				log.Printf("[watchdog][Monitor] Monitoring serviceName=%s", service.Name)
-				result := service.EvaluateConditions()
-				metric.PublishMetricsForService(service, result)
-				rwLock.Lock()
-				serviceResults[service.Name] = append(serviceResults[service.Name], result)
-				if len(serviceResults[service.Name]) > 10 {
-					serviceResults[service.Name] = serviceResults[service.Name][1:]
-				}
-				rwLock.Unlock()
-				log.Printf(
-					"[watchdog][Monitor] Finished monitoring serviceName=%s; errors=%d; requestDuration=%s",
-					service.Name,
-					len(result.Errors),
-					result.Duration.Round(time.Millisecond),
-				)
-				log.Printf("[watchdog][Monitor] Waiting interval=%s before monitoring serviceName=%s", service.Interval, service.Name)
-				time.Sleep(service.Interval)
-			}
-		}(service)
-		// To prevent multiple requests from running exactly at the same time
-		time.Sleep(100 * time.Millisecond)
+		go monitor(service)
+		// To prevent multiple requests from running at the same time
+		time.Sleep(500 * time.Millisecond)
+	}
+}
+
+func monitor(service *core.Service) {
+	for {
+		// By placing the lock here, we prevent multiple services from being monitored at the exact same time, which
+		// could cause performance issues and return inaccurate results
+		rwLock.Lock()
+		log.Printf("[watchdog][Monitor] Monitoring serviceName=%s", service.Name)
+		result := service.EvaluateConditions()
+		metric.PublishMetricsForService(service, result)
+		serviceResults[service.Name] = append(serviceResults[service.Name], result)
+		if len(serviceResults[service.Name]) > 20 {
+			serviceResults[service.Name] = serviceResults[service.Name][1:]
+		}
+		rwLock.Unlock()
+		log.Printf(
+			"[watchdog][Monitor] Finished monitoring serviceName=%s; errors=%d; requestDuration=%s",
+			service.Name,
+			len(result.Errors),
+			result.Duration.Round(time.Millisecond),
+		)
+		log.Printf("[watchdog][Monitor] Waiting interval=%s before monitoring serviceName=%s", service.Interval, service.Name)
+		time.Sleep(service.Interval)
 	}
 }