Add health check endpoint (#18465)

* chore: add health check endpoint

docs: update document about health check

fix: fix up Sqlite3 ping. current ping will success even if the db file is missing

fix: do not expose privacy information in output field

* refactor: remove HealthChecker struct

* Added `/api/healthz` to install routes.

This was needed for using /api/healthz endpoint in Docker healthchecks,
otherwise, Docker would never become healthy if using healthz endpoint
and users would not be able to complete the installation of Gitea.

* Update modules/cache/cache.go

* fine tune

* Remove unnecessary test code. Now there are 2 routes for installation (and maybe more in future)

Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: 6543 <6543@obermui.de>
Co-authored-by: Marcos de Oliveira <marcossantos@furb.br>

7 files changed, 261 insertions, 13 deletions

diff --git a/docs/content/doc/installation/on-kubernetes.en-us.md b/docs/content/doc/installation/on-kubernetes.en-us.md
index 9fe869254c..abfbdf1679 100644
--- a/docs/content/doc/installation/on-kubernetes.en-us.md
+++ b/docs/content/doc/installation/on-kubernetes.en-us.md
@@ -25,3 +25,47 @@ helm install gitea gitea-charts/gitea
 ```
 
 If you would like to customize your install, which includes kubernetes ingress, please refer to the complete [Gitea helm chart configuration details](https://gitea.com/gitea/helm-chart/)
+
+## Health check endpoint
+
+Gitea comes with a health check endpoint `/api/healthz`, you can configure it in kubernetes like this:
+
+```yaml
+  livenessProbe:
+    httpGet:
+      path: /api/healthz
+      port: http
+    initialDelaySeconds: 200
+    timeoutSeconds: 5
+    periodSeconds: 10
+    successThreshold: 1
+    failureThreshold: 10
+```
+
+a successful health check response will respond with http code `200`, here's example:
+
+```
+HTTP/1.1 200 OK
+
+
+{
+  "status": "pass",
+  "description": "Gitea: Git with a cup of tea",
+  "checks": {
+    "cache:ping": [
+      {
+        "status": "pass",
+        "time": "2022-02-19T09:16:08Z"
+      }
+    ],
+    "database:ping": [
+      {
+        "status": "pass",
+        "time": "2022-02-19T09:16:08Z"
+      }
+    ]
+  }
+}
+```
+
+for more information, please reference to kubernetes documentation [Define a liveness HTTP request](https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/#define-a-liveness-http-request)
diff --git a/docs/content/doc/installation/on-kubernetes.zh-tw.md b/docs/content/doc/installation/on-kubernetes.zh-tw.md
index 9add5c4ee1..5ea412aa00 100644
--- a/docs/content/doc/installation/on-kubernetes.zh-tw.md
+++ b/docs/content/doc/installation/on-kubernetes.zh-tw.md
@@ -25,3 +25,47 @@ helm install gitea gitea-charts/gitea
 ```
 
 若您想自訂安裝（包括使用 kubernetes ingress），請前往完整的 [Gitea helm chart configuration details](https://gitea.com/gitea/helm-chart/)
+
+##運行狀況檢查終端節點
+
+Gitea 附帶了一個運行狀況檢查端點 `/api/healthz`，你可以像這樣在 kubernetes 中配置它:
+
+```yaml
+  livenessProbe:
+    httpGet:
+      path: /api/healthz
+      port: http
+    initialDelaySeconds: 200
+    timeoutSeconds: 5
+    periodSeconds: 10
+    successThreshold: 1
+    failureThreshold: 10
+```
+
+成功的運行狀況檢查回應將使用 HTTP 代碼 `200` 進行回應，下面是示例:
+
+```
+HTTP/1.1 200 OK
+
+
+{
+  "status": "pass",
+  "description": "Gitea: Git with a cup of tea",
+  "checks": {
+    "cache:ping": [
+      {
+        "status": "pass",
+        "time": "2022-02-19T09:16:08Z"
+      }
+    ],
+    "database:ping": [
+      {
+        "status": "pass",
+        "time": "2022-02-19T09:16:08Z"
+      }
+    ]
+  }
+}
+```
+
+有關更多信息，請參考kubernetes文檔[定義一個存活態 HTTP請求接口]（https://kubernetes.io/zh/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/）
diff --git a/modules/cache/cache.go b/modules/cache/cache.go
index 0198f8da73..fd32aa153b 100644
--- a/modules/cache/cache.go
+++ b/modules/cache/cache.go
@@ -5,6 +5,7 @@
 package cache
 
 import (
+	"errors"
 	"fmt"
 	"strconv"
 
@@ -34,25 +35,37 @@ func NewContext() error {
 		if conn, err = newCache(setting.CacheService.Cache); err != nil {
 			return err
 		}
-		const testKey = "__gitea_cache_test"
-		const testVal = "test-value"
-		if err = conn.Put(testKey, testVal, 10); err != nil {
+		if err = Ping(); err != nil {
 			return err
 		}
-		val := conn.Get(testKey)
-		if valStr, ok := val.(string); !ok || valStr != testVal {
-			// If the cache is full, the Get may not read the expected value stored by Put.
-			// Since we have checked that Put can success, so we just show a warning here, do not return an error to panic.
-			log.Warn("cache (adapter:%s, config:%s) doesn't seem to work correctly, set test value '%v' but get '%v'",
-				setting.CacheService.Cache.Adapter, setting.CacheService.Cache.Conn,
-				testVal, val,
-			)
-		}
 	}
 
 	return err
 }
 
+// Ping checks if the cache service works or not, it not, it returns an error
+func Ping() error {
+	if conn == nil {
+		return errors.New("cache not available")
+	}
+	var err error
+	const testKey = "__gitea_cache_test"
+	const testVal = "test-value"
+	if err = conn.Put(testKey, testVal, 10); err != nil {
+		return err
+	}
+	val := conn.Get(testKey)
+	if valStr, ok := val.(string); !ok || valStr != testVal {
+		// If the cache is full, the Get may not read the expected value stored by Put.
+		// Since we have checked that Put can success, so we just show a warning here, do not return an error to panic.
+		log.Warn("cache (adapter:%s, config:%s) doesn't seem to work correctly, set test value '%v' but get '%v'",
+			setting.CacheService.Cache.Adapter, setting.CacheService.Cache.Conn,
+			testVal, val,
+		)
+	}
+	return nil
+}
+
 // GetCache returns the currently configured cache
 func GetCache() mc.Cache {
 	return conn
diff --git a/routers/install/routes.go b/routers/install/routes.go
index ef96e99628..e77081afe0 100644
--- a/routers/install/routes.go
+++ b/routers/install/routes.go
@@ -16,6 +16,7 @@ import (
 	"code.gitea.io/gitea/modules/web"
 	"code.gitea.io/gitea/modules/web/middleware"
 	"code.gitea.io/gitea/routers/common"
+	"code.gitea.io/gitea/routers/web/healthcheck"
 	"code.gitea.io/gitea/services/forms"
 
 	"gitea.com/go-chi/session"
@@ -106,6 +107,7 @@ func Routes() *web.Route {
 	r.Use(Init)
 	r.Get("/", Install)
 	r.Post("/", web.Bind(forms.InstallForm{}), SubmitInstall)
+	r.Get("/api/healthz", healthcheck.Check)
 
 	r.NotFound(web.Wrap(installNotFound))
 	return r
diff --git a/routers/install/routes_test.go b/routers/install/routes_test.go
index 35a66c1c47..29003c3841 100644
--- a/routers/install/routes_test.go
+++ b/routers/install/routes_test.go
@@ -13,7 +13,6 @@ import (
 func TestRoutes(t *testing.T) {
 	routes := Routes()
 	assert.NotNil(t, routes)
-	assert.Len(t, routes.R.Routes(), 1)
 	assert.EqualValues(t, "/", routes.R.Routes()[0].Pattern)
 	assert.Nil(t, routes.R.Routes()[0].SubRoutes)
 	assert.Len(t, routes.R.Routes()[0].Handlers, 2)
diff --git a/routers/web/healthcheck/check.go b/routers/web/healthcheck/check.go
new file mode 100644
index 0000000000..481f05c0da
--- /dev/null
+++ b/routers/web/healthcheck/check.go
@@ -0,0 +1,143 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package healthcheck
+
+import (
+	"net/http"
+	"os"
+	"time"
+
+	"code.gitea.io/gitea/models/db"
+	"code.gitea.io/gitea/modules/cache"
+	"code.gitea.io/gitea/modules/json"
+	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/setting"
+)
+
+type status string
+
+const (
+	// pass healthy (acceptable aliases: "ok" to support Node's Terminus and "up" for Java's SpringBoot)
+	// fail unhealthy (acceptable aliases: "error" to support Node's Terminus and "down" for Java's SpringBoot), and
+	// warn healthy, with some concerns.
+	//
+	// ref https://datatracker.ietf.org/doc/html/draft-inadarei-api-health-check#section-3.1
+	// status: (required) indicates whether the service status is acceptable
+	// or not.  API publishers SHOULD use following values for the field:
+	// The value of the status field is case-insensitive and is tightly
+	// related with the HTTP response code returned by the health endpoint.
+	// For "pass" status, HTTP response code in the 2xx-3xx range MUST be
+	// used.  For "fail" status, HTTP response code in the 4xx-5xx range
+	// MUST be used.  In case of the "warn" status, endpoints MUST return
+	// HTTP status in the 2xx-3xx range, and additional information SHOULD
+	// be provided, utilizing optional fields of the response.
+	pass status = "pass"
+	fail status = "fail"
+	warn status = "warn"
+)
+
+func (s status) ToHTTPStatus() int {
+	if s == pass || s == warn {
+		return http.StatusOK
+	}
+	return http.StatusFailedDependency
+}
+
+type checks map[string][]componentStatus
+
+// response is the data returned by the health endpoint, which will be marshaled to JSON format
+type response struct {
+	Status      status `json:"status"`
+	Description string `json:"description"`      // a human-friendly description of the service
+	Checks      checks `json:"checks,omitempty"` // The Checks Object, should be omitted on installation route
+}
+
+// componentStatus presents one status of a single check object
+// an object that provides detailed health statuses of additional downstream systems and endpoints
+// which can affect the overall health of the main API.
+type componentStatus struct {
+	Status status `json:"status"`
+	Time   string `json:"time"`             // the date-time, in ISO8601 format
+	Output string `json:"output,omitempty"` // this field SHOULD be omitted for "pass" state.
+}
+
+// Check is the health check API handler
+func Check(w http.ResponseWriter, r *http.Request) {
+	rsp := response{
+		Status:      pass,
+		Description: setting.AppName,
+		Checks:      make(checks),
+	}
+
+	statuses := make([]status, 0)
+	if setting.InstallLock {
+		statuses = append(statuses, checkDatabase(rsp.Checks))
+		statuses = append(statuses, checkCache(rsp.Checks))
+	}
+	for _, s := range statuses {
+		if s != pass {
+			rsp.Status = fail
+			break
+		}
+	}
+
+	data, _ := json.MarshalIndent(rsp, "", "  ")
+	w.Header().Set("Content-Type", "application/json")
+	w.WriteHeader(rsp.Status.ToHTTPStatus())
+	_, _ = w.Write(data)
+}
+
+// database checks gitea database status
+func checkDatabase(checks checks) status {
+	st := componentStatus{}
+	if err := db.GetEngine(db.DefaultContext).Ping(); err != nil {
+		st.Status = fail
+		st.Time = getCheckTime()
+		log.Error("database ping failed with error: %v", err)
+	} else {
+		st.Status = pass
+		st.Time = getCheckTime()
+	}
+
+	if setting.Database.UseSQLite3 && st.Status == pass {
+		if !setting.EnableSQLite3 {
+			st.Status = fail
+			st.Time = getCheckTime()
+			log.Error("SQLite3 health check failed with error: %v", "this Gitea binary is built without SQLite3 enabled")
+		} else {
+			if _, err := os.Stat(setting.Database.Path); err != nil {
+				st.Status = fail
+				st.Time = getCheckTime()
+				log.Error("SQLite3 file exists check failed with error: %v", err)
+			}
+		}
+	}
+
+	checks["database:ping"] = []componentStatus{st}
+	return st.Status
+}
+
+// cache checks gitea cache status
+func checkCache(checks checks) status {
+	if !setting.CacheService.Enabled {
+		return pass
+	}
+
+	st := componentStatus{}
+	if err := cache.Ping(); err != nil {
+		st.Status = fail
+		st.Time = getCheckTime()
+		log.Error("cache ping failed with error: %v", err)
+	} else {
+		st.Status = pass
+		st.Time = getCheckTime()
+	}
+	checks["cache:ping"] = []componentStatus{st}
+	return st.Status
+}
+
+func getCheckTime() string {
+	return time.Now().UTC().Format(time.RFC3339)
+}
diff --git a/routers/web/web.go b/routers/web/web.go
index 22b8e7cdf3..dcaad3d2bd 100644
--- a/routers/web/web.go
+++ b/routers/web/web.go
@@ -31,6 +31,7 @@ import (
 	"code.gitea.io/gitea/routers/web/events"
 	"code.gitea.io/gitea/routers/web/explore"
 	"code.gitea.io/gitea/routers/web/feed"
+	"code.gitea.io/gitea/routers/web/healthcheck"
 	"code.gitea.io/gitea/routers/web/misc"
 	"code.gitea.io/gitea/routers/web/org"
 	"code.gitea.io/gitea/routers/web/repo"
@@ -191,6 +192,8 @@ func Routes() *web.Route {
 		rw.WriteHeader(http.StatusOK)
 	})
 
+	routes.Get("/api/healthz", healthcheck.Check)
+
 	// Removed: toolbox.Toolboxer middleware will provide debug information which seems unnecessary
 	common = append(common, context.Contexter())