feat(maintenance): Per-endpoint maintenance configuration (#982)

* feat: add endpoint.maintenance-windows array for per-endpoint maintenance configuration * doc: initial entry for maintenance windows in endpoint config * doc: example documentation for per-endpoint configuration of maintenance windows * chore: var => := * test: add checks for maintenance window defaults in endpoint configuration * chore: clean up new-lines --------- Co-authored-by: TwiN <[email protected]>
TwiN · Feb 17, 2025 · a1f7bd7 · a1f7bd7
1 parent 7e122a9
commit a1f7bd7
Show file tree

Hide file tree

Showing 4 changed files with 46 additions and 5 deletions.
diff --git a/README.md b/README.md
@@ -273,6 +273,7 @@ You can then configure alerts to be triggered when an endpoint is unhealthy once
 | `endpoints[].ssh.username`                      | SSH username (e.g. example).                                                                                                                | Required `""`              |
 | `endpoints[].ssh.password`                      | SSH password (e.g. password).                                                                                                               | Required `""`              |
 | `endpoints[].alerts`                            | List of all alerts for a given endpoint. <br />See [Alerting](#alerting).                                                                   | `[]`                       |
+| `endpoints[].maintenance-windows`               | List of all maintenance windows for a given endpoint. <br />See [Maintenance](#maintenance).                                                | `[]`                       |
 | `endpoints[].client`                            | [Client configuration](#client-configuration).                                                                                              | `{}`                       |
 | `endpoints[].ui`                                | UI configuration at the endpoint level.                                                                                                     | `{}`                       |
 | `endpoints[].ui.hide-conditions`                | Whether to hide conditions from the results. Note that this only hides conditions from results evaluated from the moment this was enabled.  | `false`                    |
@@ -1710,6 +1711,19 @@ maintenance:
     - Monday
     - Thursday
 ```
+You can also specify maintenance windows on a per-endpoint basis:
+```yaml
+endpoints:
+  - name: endpoint-1
+    url: "https://example.org"
+    maintenance-windows:
+      - start: "07:30"
+        duration: 40m
+        timezone: "Europe/Berlin"
+      - start: "14:30"
+        duration: 1h
+        timezone: "Europe/Berlin"
+```
 
 
 ### Security

diff --git a/config/endpoint/endpoint.go b/config/endpoint/endpoint.go
@@ -13,6 +13,7 @@ import (
 	"strings"
 	"time"
 
+	"github.com/TwiN/gatus/v5/config/maintenance"
 	"github.com/TwiN/gatus/v5/alerting/alert"
 	"github.com/TwiN/gatus/v5/client"
 	"github.com/TwiN/gatus/v5/config/endpoint/dns"
@@ -104,6 +105,9 @@ type Endpoint struct {
 	// Alerts is the alerting configuration for the endpoint in case of failure
 	Alerts []*alert.Alert `yaml:"alerts,omitempty"`
 
+	// MaintenanceWindow is the configuration for per-endpoint maintenance windows
+	MaintenanceWindows []*maintenance.Config `yaml:"maintenance-windows,omitempty"`
+
 	// DNSConfig is the configuration for DNS monitoring
 	DNSConfig *dns.Config `yaml:"dns,omitempty"`
 
@@ -219,6 +223,11 @@ func (e *Endpoint) ValidateAndSetDefaults() error {
 	if e.Type() == TypeUNKNOWN {
 		return ErrUnknownEndpointType
 	}
+	for _, maintenanceWindow := range e.MaintenanceWindows {
+		if err := maintenanceWindow.ValidateAndSetDefaults(); err != nil {
+			return err
+		}
+	}
 	// Make sure that the request can be created
 	_, err := http.NewRequest(e.Method, e.URL, bytes.NewBuffer([]byte(e.Body)))
 	if err != nil {

diff --git a/config/endpoint/endpoint_test.go b/config/endpoint/endpoint_test.go
@@ -16,6 +16,7 @@ import (
 	"github.com/TwiN/gatus/v5/config/endpoint/dns"
 	"github.com/TwiN/gatus/v5/config/endpoint/ssh"
 	"github.com/TwiN/gatus/v5/config/endpoint/ui"
+	"github.com/TwiN/gatus/v5/config/maintenance"
 	"github.com/TwiN/gatus/v5/test"
 )
 
@@ -390,10 +391,11 @@ func TestEndpoint_Type(t *testing.T) {
 
 func TestEndpoint_ValidateAndSetDefaults(t *testing.T) {
 	endpoint := Endpoint{
-		Name:       "website-health",
-		URL:        "https://twin.sh/health",
-		Conditions: []Condition{Condition("[STATUS] == 200")},
-		Alerts:     []*alert.Alert{{Type: alert.TypePagerDuty}},
+		Name:               "website-health",
+		URL:                "https://twin.sh/health",
+		Conditions:         []Condition{Condition("[STATUS] == 200")},
+		Alerts:             []*alert.Alert{{Type: alert.TypePagerDuty}},
+		MaintenanceWindows: []*maintenance.Config{{Start: "03:50", Duration: 4 * time.Hour}},
 	}
 	if err := endpoint.ValidateAndSetDefaults(); err != nil {
 		t.Errorf("Expected no error, got %v", err)
@@ -432,6 +434,15 @@ func TestEndpoint_ValidateAndSetDefaults(t *testing.T) {
 	if endpoint.Alerts[0].FailureThreshold != 3 {
 		t.Error("Endpoint alert should've defaulted to a failure threshold of 3")
 	}
+	if len(endpoint.MaintenanceWindows) != 1 {
+		t.Error("Endpoint should've had 1 maintenance window")
+	}
+	if !endpoint.MaintenanceWindows[0].IsEnabled() {
+		t.Error("Endpoint maintenance should've defaulted to true")
+	}
+	if endpoint.MaintenanceWindows[0].Timezone != "UTC" {
+		t.Error("Endpoint maintenance should've defaulted to UTC")
+	}
 }
 
 func TestEndpoint_ValidateAndSetDefaultsWithInvalidCondition(t *testing.T) {

diff --git a/watchdog/watchdog.go b/watchdog/watchdog.go
@@ -80,7 +80,14 @@ func execute(ep *endpoint.Endpoint, alertingConfig *alerting.Config, maintenance
 	} else {
 		logr.Infof("[watchdog.execute] Monitored group=%s; endpoint=%s; key=%s; success=%v; errors=%d; duration=%s", ep.Group, ep.Name, ep.Key(), result.Success, len(result.Errors), result.Duration.Round(time.Millisecond))
 	}
-	if !maintenanceConfig.IsUnderMaintenance() {
+	inEndpointMaintenanceWindow := false
+	for _, maintenanceWindow := range ep.MaintenanceWindows {
+		if maintenanceWindow.IsUnderMaintenance() {
+			logr.Debug("[watchdog.execute] Under endpoint maintenance window")
+			inEndpointMaintenanceWindow = true
+		}
+	}
+	if !maintenanceConfig.IsUnderMaintenance() && !inEndpointMaintenanceWindow {
 		// TODO: Consider moving this after the monitoring lock is unlocked? I mean, how much noise can a single alerting provider cause...
 		HandleAlerting(ep, result, alertingConfig)
 	} else {