check: add HTTP/HTTPS check type with per-URL line graphs

.air.toml

@@ -3,7 +3,7 @@ testdata_dir = "testdata"
 tmp_dir = "tmp"
 
 [build]
-  entrypoint = ["out/wasgehtd", "-log-level=debug"]
+  entrypoint = ["out/wasgehtd"]
   cmd = "make build"
   delay = 1000
   exclude_dir = ["data"]

README.md

@@ -11,8 +11,9 @@
 - **Extensible Check System**: Modular check types via a Registry/Factory pattern. Each check type implements a common `Check` interface and declares its own metrics through a `Descriptor`.
 - **Built-in Check Types**:
   - **ping**: ICMP echo requests for host availability and latency.
+  - **http**: HTTP/HTTPS endpoint reachability and per-URL response time.
   - **wifi_stations**: Scrapes a Prometheus metrics endpoint for connected WiFi client counts per radio interface.
-- **Multi-Metric Checks**: Checks can produce multiple metrics stored as separate data sources in a single RRD file. Multi-metric checks render as stacked area graphs.
+- **Multi-Metric Checks**: Checks can produce multiple metrics stored as separate data sources in a single RRD file. Multi-metric checks render as stacked area graphs or colored line graphs depending on the check type.
 - **Host Status Aggregation**: Each host has an aggregate status (`up`, `down`, `degraded`, `unknown`) computed from all its checks. A check must be alive and have reported within the last 5 minutes to count as healthy.
 - **RRD Storage**: Uses Round Robin Databases for time-series data, with configurable archives from 1-minute resolution (1 week) to 8-hour resolution (5 years).
 - **Graph Generation**: Generates historical graphs at multiple time scales (15 minutes through 5 years) for each check type on each host.
@@ -115,7 +116,10 @@ Hosts are defined in a JSON file. Each host can specify an address and a set of
 	"google": {
 		"address": "8.8.8.8",
 		"checks": {
-			"ping": { "timeout": "5s" }
+			"ping": { "timeout": "5s" },
+			"http": {
+				"urls": ["https://www.google.com"]
+			}
 		}
 	},
 	"ap1": {
@@ -126,6 +130,19 @@ Hosts are defined in a JSON file. Each host can specify an address and a set of
 			}
 		}
 	},
+	"qube": {
+		"checks": {
+			"ping": {},
+			"http": {
+				"urls": [
+					"http://qube.example.com:2018/sign.json",
+					"https://whatsup.example.com",
+					"http://mrtg.example.com"
+				],
+				"timeout": "15s"
+			}
+		}
+	},
 	"disabled-example": {
 		"checks": {
 			"ping": { "enabled": false }
@@ -146,6 +163,19 @@ Sends ICMP echo requests to check host availability and measure latency.
 | `count`   | number | `1`     | Number of ping packets to send |
 | `enabled` | bool   | `true`  | Set to `false` to disable      |
 
+#### http
+
+Performs HTTP GET requests to a list of URLs and reports per-URL response time. Each URL becomes a separate data source in the RRD, rendered as colored lines on the graph. The check succeeds only if all configured URLs return a response (any HTTP status code counts as reachable). Redirects are not followed.
+
+TLS certificate verification is skipped by default to support locally signed certificates.
+
+| Option        | Type     | Default      | Description                        |
+| ------------- | -------- | ------------ | ---------------------------------- |
+| `urls`        | []string | _(required)_ | List of full URLs to check         |
+| `timeout`     | string   | `"10s"`      | HTTP request timeout (Go duration) |
+| `skip_verify` | bool     | `true`       | Skip TLS certificate verification  |
+| `enabled`     | bool     | `true`       | Set to `false` to disable          |
+
 #### wifi_stations
 
 Scrapes a Prometheus metrics endpoint for `wifi_stations{ifname="..."}` gauge values, reporting connected client counts per radio interface. Each configured radio becomes a separate data source in the RRD, rendered as a stacked area graph.
@@ -195,6 +225,13 @@ Returns JSON with the status of all hosts:
 					"latency_us": 12345
 				},
 				"lastupdate": 1700000000
+			},
+			"http": {
+				"alive": true,
+				"metrics": {
+					"https://www.google.com": 45230
+				},
+				"lastupdate": 1700000000
 			}
 		}
 	},
@@ -248,7 +285,8 @@ data/
 │   ├── router/
 │   │   └── ping.rrd
 │   ├── google/
-│   │   └── ping.rrd
+│   │   ├── ping.rrd
+│   │   └── http.rrd
 │   ├── ap1/
 │   │   ├── ping.rrd
 │   │   └── wifi_stations.rrd
@@ -259,6 +297,10 @@ data/
         │   ├── router_ping_15m.png
         │   ├── router_ping_1h.png
         │   └── ...
+        ├── google/
+        │   ├── google_ping_15m.png
+        │   ├── google_http_15m.png
+        │   └── ...
         ├── ap1/
         │   ├── ap1_ping_15m.png
         │   ├── ap1_ping_1h.png
@@ -268,7 +310,7 @@ data/
         └── ...
 ```
 
-Each check type gets its own RRD file (e.g., `ping.rrd`, `wifi_stations.rrd`). Multi-metric checks store all their data sources in a single RRD file.
+Each check type gets its own RRD file (e.g., `ping.rrd`, `http.rrd`, `wifi_stations.rrd`). Multi-metric checks store all their data sources in a single RRD file.
 
 ## Makefile Targets
 

pkg/check/descriptor.go

@@ -1,5 +1,19 @@
 package check
 
+// Graph style constants for Descriptor.GraphStyle.
+const (
+	// GraphStyleStack renders multi-metric graphs as stacked areas
+	// (AREA for first metric, STACK for subsequent). This is the default
+	// when GraphStyle is empty.
+	GraphStyleStack = "stack"
+
+	// GraphStyleLine renders multi-metric graphs as colored lines
+	// (LINE2 for each metric). Useful for check types where metrics
+	// are independent measurements (e.g. per-URL response times)
+	// rather than parts of a whole.
+	GraphStyleLine = "line"
+)
+
 // MetricDef describes a single metric produced by a check type.
 type MetricDef struct {
 	// ResultKey is the key used in Result.Metrics (e.g. "latency_us").
@@ -26,6 +40,17 @@ type MetricDef struct {
 // via Check.Describe(), allowing config-dependent metric shapes (e.g.
 // a wifi_stations check with a variable number of radios).
 type Descriptor struct {
+	// GraphStyle controls how multi-metric graphs are rendered.
+	// Empty or "stack" means stacked areas (default). "line" means
+	// colored lines. Single-metric checks ignore this field.
+	GraphStyle string
+
+	// Label is a human-readable label used for graph titles and the
+	// vertical axis. If empty, the first metric's Label is used.
+	// Useful when individual metric labels are long or not suitable
+	// for titles (e.g. full URLs).
+	Label string
+
 	// Metrics lists the metrics this check instance produces.
 	Metrics []MetricDef
 }

pkg/check/descriptor_test.go

@@ -9,6 +9,9 @@ func TestDescriptor_ZeroValue(t *testing.T) {
 	if d.Metrics != nil {
 		t.Error("zero Descriptor should have nil Metrics")
 	}
+	if d.GraphStyle != "" {
+		t.Error("zero Descriptor should have empty GraphStyle")
+	}
 }
 
 func TestDescriptor_WithMetrics(t *testing.T) {
@@ -45,3 +48,42 @@ func TestDescriptor_ScaleOneMeansNoScaling(t *testing.T) {
 		t.Errorf("expected Scale 1, got %d", d.Scale)
 	}
 }
+
+func TestDescriptor_GraphStyleStack(t *testing.T) {
+	d := Descriptor{
+		GraphStyle: GraphStyleStack,
+		Metrics:    []MetricDef{{ResultKey: "a", DSName: "a", Label: "a", Unit: "x"}},
+	}
+	if d.GraphStyle != "stack" {
+		t.Errorf("expected GraphStyle 'stack', got %q", d.GraphStyle)
+	}
+}
+
+func TestDescriptor_GraphStyleLine(t *testing.T) {
+	d := Descriptor{
+		GraphStyle: GraphStyleLine,
+		Metrics:    []MetricDef{{ResultKey: "a", DSName: "a", Label: "a", Unit: "x"}},
+	}
+	if d.GraphStyle != "line" {
+		t.Errorf("expected GraphStyle 'line', got %q", d.GraphStyle)
+	}
+}
+
+func TestDescriptor_EmptyGraphStyleDefaultsToStack(t *testing.T) {
+	// Empty string should be treated as stack by the graph layer
+	d := Descriptor{
+		Metrics: []MetricDef{{ResultKey: "a", DSName: "a", Label: "a", Unit: "x"}},
+	}
+	if d.GraphStyle != "" {
+		t.Errorf("expected empty GraphStyle, got %q", d.GraphStyle)
+	}
+}
+
+func TestGraphStyleConstants(t *testing.T) {
+	if GraphStyleStack != "stack" {
+		t.Errorf("expected GraphStyleStack='stack', got %q", GraphStyleStack)
+	}
+	if GraphStyleLine != "line" {
+		t.Errorf("expected GraphStyleLine='line', got %q", GraphStyleLine)
+	}
+}