khenaidoo | d948f77 | 2021-08-11 17:49:24 -0400 | [diff] [blame] | 1 | // Copyright 2017 The etcd Authors |
| 2 | // |
| 3 | // Licensed under the Apache License, Version 2.0 (the "License"); |
| 4 | // you may not use this file except in compliance with the License. |
| 5 | // You may obtain a copy of the License at |
| 6 | // |
| 7 | // http://www.apache.org/licenses/LICENSE-2.0 |
| 8 | // |
| 9 | // Unless required by applicable law or agreed to in writing, software |
| 10 | // distributed under the License is distributed on an "AS IS" BASIS, |
| 11 | // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 12 | // See the License for the specific language governing permissions and |
| 13 | // limitations under the License. |
| 14 | |
| 15 | package etcdhttp |
| 16 | |
| 17 | import ( |
| 18 | "context" |
| 19 | "encoding/json" |
| 20 | "net/http" |
| 21 | "time" |
| 22 | |
| 23 | "github.com/coreos/etcd/etcdserver" |
| 24 | "github.com/coreos/etcd/etcdserver/etcdserverpb" |
| 25 | "github.com/coreos/etcd/raft" |
| 26 | |
| 27 | "github.com/prometheus/client_golang/prometheus" |
| 28 | "github.com/prometheus/client_golang/prometheus/promhttp" |
| 29 | ) |
| 30 | |
| 31 | const ( |
| 32 | PathMetrics = "/metrics" |
| 33 | PathHealth = "/health" |
| 34 | ) |
| 35 | |
| 36 | // HandleMetricsHealth registers metrics and health handlers. |
| 37 | func HandleMetricsHealth(mux *http.ServeMux, srv etcdserver.ServerV2) { |
| 38 | mux.Handle(PathMetrics, promhttp.Handler()) |
| 39 | mux.Handle(PathHealth, NewHealthHandler(func() Health { return checkHealth(srv) })) |
| 40 | } |
| 41 | |
| 42 | // HandlePrometheus registers prometheus handler on '/metrics'. |
| 43 | func HandlePrometheus(mux *http.ServeMux) { |
| 44 | mux.Handle(PathMetrics, promhttp.Handler()) |
| 45 | } |
| 46 | |
| 47 | // NewHealthHandler handles '/health' requests. |
| 48 | func NewHealthHandler(hfunc func() Health) http.HandlerFunc { |
| 49 | return func(w http.ResponseWriter, r *http.Request) { |
| 50 | if r.Method != http.MethodGet { |
| 51 | w.Header().Set("Allow", http.MethodGet) |
| 52 | http.Error(w, "Method Not Allowed", http.StatusMethodNotAllowed) |
| 53 | plog.Warningf("/health error (status code %d)", http.StatusMethodNotAllowed) |
| 54 | return |
| 55 | } |
| 56 | h := hfunc() |
| 57 | d, _ := json.Marshal(h) |
| 58 | if h.Health != "true" { |
| 59 | http.Error(w, string(d), http.StatusServiceUnavailable) |
| 60 | return |
| 61 | } |
| 62 | w.WriteHeader(http.StatusOK) |
| 63 | w.Write(d) |
| 64 | } |
| 65 | } |
| 66 | |
| 67 | var ( |
| 68 | healthSuccess = prometheus.NewCounter(prometheus.CounterOpts{ |
| 69 | Namespace: "etcd", |
| 70 | Subsystem: "server", |
| 71 | Name: "health_success", |
| 72 | Help: "The total number of successful health checks", |
| 73 | }) |
| 74 | healthFailed = prometheus.NewCounter(prometheus.CounterOpts{ |
| 75 | Namespace: "etcd", |
| 76 | Subsystem: "server", |
| 77 | Name: "health_failures", |
| 78 | Help: "The total number of failed health checks", |
| 79 | }) |
| 80 | ) |
| 81 | |
| 82 | func init() { |
| 83 | prometheus.MustRegister(healthSuccess) |
| 84 | prometheus.MustRegister(healthFailed) |
| 85 | } |
| 86 | |
| 87 | // Health defines etcd server health status. |
| 88 | // TODO: remove manual parsing in etcdctl cluster-health |
| 89 | type Health struct { |
| 90 | Health string `json:"health"` |
| 91 | } |
| 92 | |
| 93 | // TODO: server NOSPACE, etcdserver.ErrNoLeader in health API |
| 94 | |
| 95 | func checkHealth(srv etcdserver.ServerV2) Health { |
| 96 | h := Health{Health: "true"} |
| 97 | |
| 98 | as := srv.Alarms() |
| 99 | if len(as) > 0 { |
| 100 | h.Health = "false" |
| 101 | for _, v := range as { |
| 102 | plog.Warningf("/health error due to an alarm %s", v.String()) |
| 103 | } |
| 104 | } |
| 105 | |
| 106 | if h.Health == "true" { |
| 107 | if uint64(srv.Leader()) == raft.None { |
| 108 | h.Health = "false" |
| 109 | plog.Warningf("/health error; no leader (status code %d)", http.StatusServiceUnavailable) |
| 110 | } |
| 111 | } |
| 112 | |
| 113 | if h.Health == "true" { |
| 114 | ctx, cancel := context.WithTimeout(context.Background(), time.Second) |
| 115 | _, err := srv.Do(ctx, etcdserverpb.Request{Method: "QGET"}) |
| 116 | cancel() |
| 117 | if err != nil { |
| 118 | h.Health = "false" |
| 119 | plog.Warningf("/health error; QGET failed %v (status code %d)", err, http.StatusServiceUnavailable) |
| 120 | } |
| 121 | } |
| 122 | |
| 123 | if h.Health == "true" { |
| 124 | healthSuccess.Inc() |
| 125 | plog.Infof("/health OK (status code %d)", http.StatusOK) |
| 126 | } else { |
| 127 | healthFailed.Inc() |
| 128 | } |
| 129 | return h |
| 130 | } |