kuma-glpi-bridge: webhook receiver opening GLPI incident tickets

Go service: Kuma webhook -> GLPI REST (two-step session) ticket per monitor down-event, dedupe while open, recovery followup + solve on up. Distroless image; CI = gofmt/vet/build/secret-scan. Ticket: https://projects.knownelement.com/issues/824
2026-09-06 12:30:48 -05:00
commit 16519fd97f
6 changed files with 281 additions and 0 deletions
@@ -0,0 +1,23 @@
+# CI [#824] — vet + build + secret scan; image builds are done by the lane
+# (digest-pinned push) per the compose/lifecycle house rules.
+name: ci
+on:
+  push:
+    branches: [main]
+  pull_request:
+jobs:
+  vet:
+    runs-on: ultix
+    container:
+      image: golang:1.23-alpine
+    steps:
+      - uses: actions/checkout@v4
+      - run: gofmt -l . | tee /tmp/fmt.out && test ! -s /tmp/fmt.out
+      - run: go vet ./...
+      - run: go build ./...
+      - name: secret scan
+        run: |
+          apk add --no-cache grep >/dev/null 2>&1 || true
+          if grep -rInE "BEGIN (RSA |OPENSSH |EC )?PRIVATE KEY|user_token [A-Za-z0-9]{20,}" --exclude-dir=.git .; then
+            echo "::error::secret material committed"; exit 1
+          fi
@@ -0,0 +1,10 @@
+FROM golang:1.23-alpine AS build
+WORKDIR /src
+COPY go.mod ./
+COPY cmd/ ./cmd/
+RUN go build -ldflags="-s -w" -o /out/bridge ./cmd/bridge
+FROM scratch
+COPY --from=build /out/bridge /bridge
+USER 65532:65532
+EXPOSE 8080
+ENTRYPOINT ["/bridge"]
@@ -0,0 +1,11 @@
+# KNELKumaGlpiBridge
+
+Uptime Kuma -> GLPI incident bridge. Kuma webhook POSTs land on
+`/webhook`; DOWN opens a GLPI incident ticket (deduped per monitor),
+UP posts a recovery followup and solves the ticket with evidence.
+
+Deployed on pfv-k8s via flux (NodePort 30801). GLPI tokens come from a
+k8s Secret created imperatively — never in git.
+
+- Redmine: https://projects.knownelement.com/issues/824
+- Discourse: incident-response category (outage flow)
@@ -0,0 +1,234 @@
+// kuma-glpi-bridge: Uptime Kuma webhook -> GLPI incident tickets.
+// Receives Kuma webhook POSTs, opens a GLPI Ticket (type 1) per monitor
+// down-event, dedupes while open, and posts a recovery followup on up.
+// Ticket: https://projects.knownelement.com/issues/824
+package main
+
+import (
+	"crypto/tls"
+	"encoding/json"
+	"fmt"
+	"log"
+	"net/http"
+	"net/url"
+	"os"
+	"strings"
+	"sync"
+	"time"
+)
+
+type kumaWebhook struct {
+	Monitor struct {
+		Name string `json:"name"`
+		URL  string `json:"url"`
+		Type string `json:"type"`
+	} `json:"monitor"`
+	Heartbeat struct {
+		Status int    `json:"status"`
+		Msg    string `json:"msg"`
+		Time   string `json:"time"`
+	} `json:"heartbeat"`
+	Msg string `json:"msg"`
+}
+
+type glpiClient struct {
+	baseURL    string
+	appToken   string
+	userToken  string
+	httpClient *http.Client
+	mu         sync.Mutex
+	session    string
+	expires    time.Time
+}
+
+func newGLPI() *glpiClient {
+	return &glpiClient{
+		baseURL:   os.Getenv("GLPI_URL"),
+		appToken:  os.Getenv("GLPI_APP_TOKEN"),
+		userToken: os.Getenv("GLPI_USER_TOKEN"),
+		httpClient: &http.Client{
+			Timeout:   20 * time.Second,
+			Transport: &http.Transport{TLSClientConfig: &tls.Config{InsecureSkipVerify: os.Getenv("GLPI_INSECURE") == "1"}}, //nolint:gosec // self-signed tailnet endpoint
+		},
+	}
+}
+
+// session returns a cached GLPI session token, re-initing when stale.
+func (g *glpiClient) ensureSession() (string, error) {
+	g.mu.Lock()
+	defer g.mu.Unlock()
+	if g.session != "" && time.Now().Before(g.expires) {
+		return g.session, nil
+	}
+	req, err := http.NewRequest("GET", g.baseURL+"/initSession", nil)
+	if err != nil {
+		return "", err
+	}
+	req.Header.Set("App-Token", g.appToken)
+	req.Header.Set("Authorization", "user_token "+g.userToken)
+	resp, err := g.httpClient.Do(req)
+	if err != nil {
+		return "", err
+	}
+	defer resp.Body.Close()
+	var out struct {
+		SessionToken string `json:"session_token"`
+	}
+	if err := json.NewDecoder(resp.Body).Decode(&out); err != nil || out.SessionToken == "" {
+		return "", fmt.Errorf("initSession failed: HTTP %d", resp.StatusCode)
+	}
+	g.session = out.SessionToken
+	g.expires = time.Now().Add(50 * time.Minute)
+	return g.session, nil
+}
+
+func (g *glpiClient) do(method, path string, body any, out any) error {
+	sess, err := g.ensureSession()
+	if err != nil {
+		return err
+	}
+	var rd *strings.Reader
+	if body != nil {
+		b, err := json.Marshal(body)
+		if err != nil {
+			return err
+		}
+		rd = strings.NewReader(string(b))
+	} else {
+		rd = strings.NewReader("")
+	}
+	req, err := http.NewRequest(method, g.baseURL+path, rd)
+	if err != nil {
+		return err
+	}
+	req.Header.Set("App-Token", g.appToken)
+	req.Header.Set("Session-Token", sess)
+	if body != nil {
+		req.Header.Set("Content-Type", "application/json")
+	}
+	resp, err := g.httpClient.Do(req)
+	if err != nil {
+		return err
+	}
+	defer resp.Body.Close()
+	if resp.StatusCode >= 300 {
+		return fmt.Errorf("%s %s: HTTP %d", method, path, resp.StatusCode)
+	}
+	if out != nil {
+		return json.NewDecoder(resp.Body).Decode(out)
+	}
+	return nil
+}
+
+// openTicketFor returns the id of an existing unsolved ticket for this monitor, if any.
+func (g *glpiClient) openTicketFor(monitor string) (int, error) {
+	q := url.Values{}
+	q.Set("is_deleted", "0")
+	q.Set("criteria[0][field]", "1") // name
+	q.Set("criteria[0][searchtype]", "contains")
+	q.Set("criteria[0][value]", "[kuma] "+monitor)
+	q.Set("criteria[1][link]", "AND")
+	q.Set("criteria[1][field]", "12") // status
+	q.Set("criteria[1][searchtype]", "equals")
+	q.Set("criteria[1][value]", "notold")
+	var out []struct {
+		ID int `json:"2"`
+	}
+	if err := g.do("GET", "/search/Ticket?"+q.Encode(), nil, &out); err != nil {
+		return 0, err
+	}
+	if len(out) > 0 {
+		return out[0].ID, nil
+	}
+	return 0, nil
+}
+
+func (g *glpiClient) createTicket(name, content string) (int, error) {
+	var out struct {
+		ID int `json:"id"`
+	}
+	err := g.do("POST", "/Ticket", map[string]any{
+		"input": map[string]any{"name": name, "content": content, "type": 1},
+	}, &out)
+	return out.ID, err
+}
+
+func (g *glpiClient) followup(ticket int, content string) error {
+	return g.do("POST", "/followup", map[string]any{
+		"input": map[string]any{"itemtype": "Ticket", "items_id": ticket, "content": content},
+	}, nil)
+}
+
+type bridge struct {
+	glpi *glpiClient
+	mu   sync.Mutex
+	open map[string]int // monitor -> ticket id
+}
+
+func (b *bridge) handle(w http.ResponseWriter, r *http.Request) {
+	if r.Method != http.MethodPost {
+		http.Error(w, "POST only", http.StatusMethodNotAllowed)
+		return
+	}
+	var k kumaWebhook
+	if err := json.NewDecoder(r.Body).Decode(&k); err != nil {
+		http.Error(w, "bad json", http.StatusBadRequest)
+		return
+	}
+	if k.Monitor.Name == "" {
+		w.WriteHeader(http.StatusAccepted)
+		return
+	}
+	status := k.Heartbeat.Status
+	name := "[kuma] " + k.Monitor.Name
+	content := fmt.Sprintf("Automated incident from Uptime Kuma bridge (#824).\nMonitor: %s (type %s)\nURL: %s\nStatus code: %d\nMessage: %s\nAt: %s\n\nSource: https://status.knownelement.com — work this incident per the incidents-first mandate; see Redmine project \"Incident Response\".",
+		k.Monitor.Name, k.Monitor.Type, k.Monitor.URL, status, k.Msg, k.Heartbeat.Time)
+
+	b.mu.Lock()
+	defer b.mu.Unlock()
+	switch status {
+	case 2: // DOWN (0=unknown,1=up,2=down per Kuma webhook)
+		if _, exists := b.open[k.Monitor.Name]; !exists {
+			tid, err := b.glpi.openTicketFor(k.Monitor.Name)
+			if err == nil && tid == 0 {
+				tid, err = b.glpi.createTicket(name, content)
+			}
+			if err != nil {
+				log.Printf("ticket create failed: %v", err)
+				http.Error(w, "glpi error", http.StatusBadGateway)
+				return
+			}
+			b.open[k.Monitor.Name] = tid
+			log.Printf("incident ticket %d for %s", tid, k.Monitor.Name)
+		}
+	case 1: // UP
+		tid, exists := b.open[k.Monitor.Name]
+		if !exists {
+			return
+		}
+		delete(b.open, k.Monitor.Name)
+		go func() {
+			fu := fmt.Sprintf("RECOVERY: monitor %s back UP at %s. Ticket auto-closing as solved with recovery evidence.", k.Monitor.Name, k.Heartbeat.Time)
+			if err := b.glpi.followup(tid, fu); err != nil {
+				log.Printf("followup failed: %v", err)
+				return
+			}
+			if err := b.glpi.do("POST", fmt.Sprintf("/Ticket/%d", tid), map[string]any{"input": map[string]any{"status": 11}}, nil); err != nil { // 11 = solved
+				log.Printf("solve failed: %v", err)
+			}
+		}()
+	}
+	w.WriteHeader(http.StatusOK)
+}
+
+func main() {
+	listen := os.Getenv("BRIDGE_LISTEN")
+	if listen == "" {
+		listen = ":8080"
+	}
+	b := &bridge{glpi: newGLPI(), open: map[string]int{}}
+	http.HandleFunc("/webhook", b.handle)
+	http.HandleFunc("/healthz", func(w http.ResponseWriter, _ *http.Request) { w.WriteHeader(200) })
+	log.Printf("kuma-glpi-bridge listening on %s", listen)
+	log.Fatal(http.ListenAndServe(listen, nil))
+}
@@ -0,0 +1,3 @@
+module git.knownelement.com/KNEL/KNELKumaGlpiBridge
+
+go 1.23