- customisable time format

- custom messages
- configure threshold count instead of uptime %

api.go

@@ -37,7 +37,7 @@ func (api CachetAPI) Ping() error {
 
 // SendMetric adds a data point to a cachet monitor
 func (api CachetAPI) SendMetric(id int, lag int64) {
-	logrus.Debugf("Sending lag metric ID:%d %vms", id, lag)
+	logrus.Debugf("Sending lag metric ID:%d RTT %vms", id, lag)
 
 	jsonBytes, _ := json.Marshal(map[string]interface{}{
 		"value":     lag,

config.go

@@ -10,13 +10,13 @@ import (
 )
 
 type CachetMonitor struct {
-	SystemName  string                   `json:"system_name"`
+	SystemName  string                   `json:"system_name" yaml:"system_name"`
+	DateFormat  string                   `json:"date_format" yaml:"date_format"`
 	API         CachetAPI                `json:"api"`
 	RawMonitors []map[string]interface{} `json:"monitors" yaml:"monitors"`
 
-	Monitors []MonitorInterface `json:"-" yaml:"-"`
+	Monitors  []MonitorInterface `json:"-" yaml:"-"`
-
+	Immediate bool               `json:"-" yaml:"-"`
-	Immediate bool `json:"-" yaml:"-"`
 }
 
 // Validate configuration
@@ -28,6 +28,10 @@ func (cfg *CachetMonitor) Validate() bool {
 		cfg.SystemName = getHostname()
 	}
 
+	if len(cfg.DateFormat) == 0 {
+		cfg.DateFormat = DefaultTimeFormat
+	}
+
 	if len(cfg.API.Token) == 0 || len(cfg.API.URL) == 0 {
 		logrus.Warnf("API URL or API Token missing.\nGet help at https://github.com/castawaylabs/cachet-monitor")
 		valid = false
@@ -74,3 +78,11 @@ func GetMonitorType(t string) string {
 
 	return t
 }
+
+func getTemplateData(monitor *AbstractMonitor) map[string]interface{} {
+	return map[string]interface{}{
+		"SystemName": monitor.config.SystemName,
+		"API":        monitor.config.API,
+		"Monitor":    monitor,
+	}
+}

http.go

@@ -8,11 +8,13 @@ import (
 	"strconv"
 	"strings"
 	"time"
+
+	"github.com/Sirupsen/logrus"
 )
 
 // Investigating template
 var defaultHTTPInvestigatingTpl = MessageTemplate{
-	Subject: `{{ .Name }} - {{ .config.SystemName }}`,
+	Subject: `{{ .Monitor.Name }} - {{ .SystemName }}`,
 	Message: `{{ .Name }} check **failed** - {{ .now }}
 
 {{ .lastFailReason }}`,
@@ -90,6 +92,7 @@ func (mon *HTTPMonitor) Validate() []string {
 	mon.Template.Fixed.SetDefault(defaultHTTPFixedTpl)
 
 	errs := mon.AbstractMonitor.Validate()
+	logrus.Warnf("%#v", mon.Template.Investigating)
 
 	if len(mon.ExpectedBody) > 0 {
 		exp, err := regexp.Compile(mon.ExpectedBody)

monitor.go

@@ -44,8 +44,9 @@ type AbstractMonitor struct {
 		Fixed         MessageTemplate
 	}
 
-	// Threshold = percentage
+	// Threshold = percentage / number of down incidents
-	Threshold float32
+	Threshold      float32
+	ThresholdCount bool `mapstructure:"threshold_count"`
 
 	history        []bool
 	lastFailReason string
@@ -83,7 +84,10 @@ func (mon *AbstractMonitor) Validate() []string {
 	}
 
 	if err := mon.Template.Fixed.Compile(); err != nil {
-		errs = append(errs, "Could not compile template: "+err.Error())
+		errs = append(errs, "Could not compile \"fixed\" template: "+err.Error())
+	}
+	if err := mon.Template.Investigating.Compile(); err != nil {
+		errs = append(errs, "Could not compile \"investigating\" template: "+err.Error())
 	}
 
 	return errs
@@ -137,11 +141,16 @@ func (mon *AbstractMonitor) tick(iface MonitorInterface) {
 	up := iface.test()
 	lag := getMs() - reqStart
 
-	if len(mon.history) == HistorySize-1 {
+	histSize := HistorySize
+	if mon.ThresholdCount {
+		histSize = int(mon.Threshold)
+	}
+
+	if len(mon.history) == histSize-1 {
 		logrus.Warnf("%v is now saturated\n", mon.Name)
 	}
-	if len(mon.history) >= HistorySize {
+	if len(mon.history) >= histSize {
-		mon.history = mon.history[len(mon.history)-(HistorySize-1):]
+		mon.history = mon.history[len(mon.history)-(histSize-1):]
 	}
 	mon.history = append(mon.history, up)
 	mon.AnalyseData()
@@ -163,43 +172,61 @@ func (monitor *AbstractMonitor) AnalyseData() {
 	}
 
 	t := (float32(numDown) / float32(len(monitor.history))) * 100
-	logrus.Printf("%s %.2f%%/%.2f%% down at %v\n", monitor.Name, t, monitor.Threshold, time.Now().UnixNano()/int64(time.Second))
+	if monitor.ThresholdCount {
+		logrus.Printf("%s %d/%d down at %v", monitor.Name, numDown, int(monitor.Threshold), time.Now().Format(DefaultTimeFormat))
+	} else {
+		logrus.Printf("%s %.2f%%/%.2f%% down at %v", monitor.Name, t, monitor.Threshold, time.Now().Format(DefaultTimeFormat))
+	}
 
-	if len(monitor.history) != HistorySize {
+	histSize := HistorySize
+	if monitor.ThresholdCount {
+		histSize = int(monitor.Threshold)
+	}
+
+	if len(monitor.history) != histSize {
 		// not saturated
 		return
 	}
 
-	if t > monitor.Threshold && monitor.incident == nil {
+	triggered := (monitor.ThresholdCount && numDown == int(monitor.Threshold)) || (!monitor.ThresholdCount && t > monitor.Threshold)
+
+	if triggered && monitor.incident == nil {
+		// create incident
+		subject, message := monitor.Template.Investigating.Exec(getTemplateData(monitor))
 		monitor.incident = &Incident{
-			Name:        monitor.Name + " - " + monitor.config.SystemName,
+			Name:        subject,
 			ComponentID: monitor.ComponentID,
-			Message:     monitor.Name + " check **failed** - " + time.Now().Format(DefaultTimeFormat),
+			Message:     message,
 			Notify:      true,
 		}
 
-		if len(monitor.lastFailReason) > 0 {
-			monitor.incident.Message += "\n\n `" + monitor.lastFailReason + "`"
-		}
-
 		// is down, create an incident
-		logrus.Printf("%v creating incident. Monitor is down: %v", monitor.Name, monitor.lastFailReason)
+		logrus.Warnf("%v: creating incident. Monitor is down: %v", monitor.Name, monitor.lastFailReason)
 		// set investigating status
 		monitor.incident.SetInvestigating()
 		// create/update incident
 		if err := monitor.incident.Send(monitor.config); err != nil {
 			logrus.Printf("Error sending incident: %v\n", err)
 		}
-	} else if t < monitor.Threshold && monitor.incident != nil {
-		// was down, created an incident, its now ok, make it resolved.
-		logrus.Printf("%v resolved downtime incident", monitor.Name)
 
-		// resolve incident
+		return
-		monitor.incident.Message = "\n**Resolved** - " + time.Now().Format(DefaultTimeFormat) + "\n\n - - - \n\n" + monitor.incident.Message
-		monitor.incident.SetFixed()
-		monitor.incident.Send(monitor.config)
-
-		monitor.lastFailReason = ""
-		monitor.incident = nil
 	}
+
+	// still triggered or no incident
+	if triggered || monitor.incident == nil {
+		return
+	}
+
+	logrus.Warnf("Resolving incident")
+
+	// was down, created an incident, its now ok, make it resolved.
+	logrus.Printf("%v resolved downtime incident", monitor.Name)
+
+	// resolve incident
+	monitor.incident.Message = "\n**Resolved** - " + time.Now().Format(DefaultTimeFormat) + "\n\n - - - \n\n" + monitor.incident.Message
+	monitor.incident.SetFixed()
+	monitor.incident.Send(monitor.config)
+
+	monitor.lastFailReason = ""
+	monitor.incident = nil
 }

template.go

@@ -1,6 +1,11 @@
 package cachet
 
-import "text/template"
+import (
+	"bytes"
+	"text/template"
+
+	"github.com/Sirupsen/logrus"
+)
 
 type MessageTemplate struct {
 	Subject string `json:"subject"`
@@ -33,6 +38,16 @@ func (t *MessageTemplate) Compile() error {
 	return err
 }
 
+func (t *MessageTemplate) Exec(data interface{}) (string, string) {
+	buf := new(bytes.Buffer)
+
+	logrus.Warnf("%#v", t.subjectTpl)
+	t.subjectTpl.Execute(buf, data)
+	subject := buf.String()
+
+	return subject, ""
+}
+
 func compileTemplate(text string) (*template.Template, error) {
 	return template.New("").Parse(text)
 }