Merge pull request #313 from weaveworks/311-scope-0.13

Update to Scope 0.13

app-mapper/main.go

@@ -19,7 +19,7 @@ import (
 )
 
 const (
-	defaultAppImage = "weaveworks/scope:0.10.0"
+	defaultAppImage = "weaveworks/scope:0.13.1"
 	defaultDBURI    = "postgres://postgres@app-mapper-db/app_mapper?sslmode=disable"
 )
 
@@ -98,13 +98,13 @@ func (f *flags) makeAppProvisioner() appProvisioner {
 
 	var p appProvisioner
 
-	args := []string{"--no-probe"}
-
+	args := []string{"--no-probe", "--app.weave.addr="}
 	switch f.appProvisioner {
 	case "docker":
 		options := dockerProvisionerOptions{
 			appConfig: docker.Config{
 				Image: f.dockerAppImage,
+				Env:   []string{"CHECKPOINT_DISABLE=true"},
 				Cmd:   args,
 			},
 			hostConfig:    docker.HostConfig{},
@@ -120,6 +120,7 @@ func (f *flags) makeAppProvisioner() appProvisioner {
 			appContainer: kapi.Container{
 				Name:  "scope",
 				Image: f.dockerAppImage,
+				Env:   []kapi.EnvVar{{Name: "CHECKPOINT_DISABLE", Value: "true"}},
 				Args:  args,
 				Ports: []kapi.ContainerPort{
 					{ContainerPort: scope.AppPort}},

app-mapper/proxy.go

@@ -51,18 +51,35 @@ func (p proxy) registerHandlers(router *mux.Router) {
 			p.forwardRequest(w, r, orgID)
 		},
 	)
-	// We believe forwards to /api/topology/{topology}/ws will normally have
-	// very high latencies. We separate that endpoint from the general
+	// We believe forwards to websocket endpoints
+	// /api/topology/{topology}/ws and /api/pipe/{pipeID} will normally have
+	// very high latencies. We separate those endpoint from the general
 	// PathPrefix in order to give it a special name, which we exclude from
 	// the latency alerting rules in our monitoring. See service #291.
 	router.Path("/api/app/{orgName}/api/topology/{topology}/ws").Name("api_app_topology_ws").Handler(fwd)
+	router.Path("/api/app/{orgName}/api/pipe/{pipeID}").Name("api_app_pipe").Handler(fwd)
 	router.PathPrefix("/api/app/{orgName}/").Name("api_app").Handler(fwd)
 
 	router.Path("/api/control/ws").Name("api_control").Handler(authProbeHandler(p.authenticator,
 		func(w http.ResponseWriter, r *http.Request, orgID string) {
 			p.forwardRequest(w, r, orgID)
 		},
 	))
+
+	// Pipe deletion from probe
+	router.Path("/api/pipe/{pipeID}").Methods("DELETE").Name("api_probe_pipe_delete").Handler(authProbeHandler(p.authenticator,
+		func(w http.ResponseWriter, r *http.Request, orgID string) {
+			p.forwardRequest(w, r, orgID)
+		},
+	))
+
+	// Pipe websocket from probe
+	router.Path("/api/pipe/{pipeID}/probe").Name("api_probe_pipe").Handler(authProbeHandler(p.authenticator,
+		func(w http.ResponseWriter, r *http.Request, orgID string) {
+			p.forwardRequest(w, r, orgID)
+		},
+	))
+
 	router.Path("/api/report").Name("api_report").Handler(authProbeHandler(p.authenticator,
 		func(w http.ResponseWriter, r *http.Request, orgID string) {
 			if probeID := r.Header.Get(scope.ScopeProbeIDHeader); probeID == "" {

client/src/pages/organization/page.jsx

@@ -124,7 +124,7 @@ export default class OrganizationPage extends React.Component {
               </div>
               <Box>
                 <div style={styles.code}>
-                  <div>sudo wget -O /usr/local/bin/scope \<br />&nbsp;&nbsp;https://github.com/weaveworks/scope/releases/download/v0.10.0/scope</div>
+                  <div>sudo wget -O /usr/local/bin/scope \<br />&nbsp;&nbsp;https://github.com/weaveworks/scope/releases/download/v0.13.1/scope</div>
                   <div>sudo chmod a+x /usr/local/bin/scope</div>
                   <div>sudo scope launch --service-token={this.state.probeToken}</div>
                 </div>

frontend/default.conf

@@ -70,8 +70,13 @@ server {
         proxy_pass http://app-mapper.default.svc.cluster.local$request_uri;
     }
 
-    # topology (from UI) and control (from probe) websockets
-    location ~ ^/api/(app/[^/]+/api/topology/[^/]+|control)/ws {
+    # pipe delete requests from probe
+    location ~ ^/api/pipe/pipe-[^/]+$ {
+        proxy_pass http://app-mapper.default.svc.cluster.local$request_uri;
+    }
+
+    # topology (from UI), pipe (from both UI and probe) and control (from probe) websockets
+    location ~ ^/api/(app/[^/]+/api/(topology/[^/]+/ws|pipe/pipe-[^/]+)|pipe/pipe-[^/]+/probe|control/ws)$ {
         proxy_pass http://app-mapper.default.svc.cluster.local$request_uri;
         proxy_http_version 1.1;
         proxy_set_header Upgrade $http_upgrade;

k8s/dev/scope/scope-app-rc.yaml

@@ -14,9 +14,13 @@ spec:
     spec:
       containers:
       - name: weave-scope-app
-        image: weaveworks/scope:0.13.0
+        image: weaveworks/scope:0.13.1
+        env:
+        - name: CHECKPOINT_DISABLE
+          value: "true"
         args:
         - --no-probe
+        - --app.weave.addr=
         - --debug
         ports:
         - containerPort: 4040

k8s/dev/scope/scope-probe-ds.yaml

@@ -17,12 +17,16 @@ spec:
       hostNetwork: true
       containers:
       - name:  weave-scope-probe
-        image: weaveworks/scope:0.13.0
+        image: weaveworks/scope:0.13.1
+        env:
+        - name: CHECKPOINT_DISABLE
+          value: "true"
         args:
         - --no-app
         - --probe.docker.bridge=cbr0
         - --probe.docker=true
         - --probe.kubernetes=true
+        - --probe.weave.router.addr=
         - --debug
 
         # We cannot refer to the Scope App with its Kubernetes Service

k8s/local/scope/scope-app-rc.yaml

@@ -14,9 +14,13 @@ spec:
     spec:
       containers:
       - name: weave-scope-app
-        image: weaveworks/scope:0.11.1
+        image: weaveworks/scope:0.13.1
+        env:
+        - name: CHECKPOINT_DISABLE
+          value: "true"
         args:
         - --no-probe
+        - --app.weave.addr=
         ports:
         - containerPort: 4040
           hostPort: 4040

k8s/local/scope/scope-probe-ds.yaml

@@ -17,11 +17,15 @@ spec:
       hostNetwork: true
       containers:
       - name:  weave-scope-probe
-        image: weaveworks/scope:0.11.1
+        image: weaveworks/scope:0.13.1
+        env:
+        - name: CHECKPOINT_DISABLE
+          value: "true"
         args:
         - --no-app
         - --probe.docker=true
         - --probe.kubernetes=true
+        - --probe.weave.router.addr=
         # We cannot refer to the Scope App with its Kubernetes Service
         # DNS name (weave-scope-app) directly because Docker doesn't
         # allow customizing the DNS servers (the Kubernetes DNS add-on

k8s/prod/scope/scope-app-rc.yaml

@@ -14,9 +14,13 @@ spec:
     spec:
       containers:
       - name: weave-scope-app
-        image: weaveworks/scope:0.13.0
+        image: weaveworks/scope:0.13.1
+        env:
+        - name: CHECKPOINT_DISABLE
+          value: "true"
         args:
         - --no-probe
+        - --app.weave.addr=
         ports:
         - containerPort: 4040
           hostPort: 4040

k8s/prod/scope/scope-probe-ds.yaml

@@ -17,12 +17,16 @@ spec:
       hostNetwork: true
       containers:
       - name:  weave-scope-probe
-        image: weaveworks/scope:0.13.0
+        image: weaveworks/scope:0.13.1
+        env:
+        - name: CHECKPOINT_DISABLE
+          value: "true"
         args:
         - --no-app
         - --probe.docker.bridge=cbr0
         - --probe.docker=true
         - --probe.kubernetes=true
+        - --probe.weave.router.addr=
 
         # We cannot refer to the Scope App with its Kubernetes Service
         # DNS name (weave-scope-app) directly because Docker doesn't

monitoring/prometheus/Dockerfile

@@ -1,4 +1,4 @@
-FROM prom/prometheus
+FROM prom/prometheus:0.17.0
 LABEL works.weave.role=system
 
 COPY prometheus.yml alert.rules /etc/prometheus/

monitoring/prometheus/alert.rules

@@ -30,7 +30,7 @@ ALERT Users4xxRate
 # https://github.com/prometheus/prometheus/issues/1289
 
 ALERT AppMapperLatency
- IF          max(sort_desc( scope_appmapper_request_duration_nanoseconds{route!="api_control", route!="api_app_topology_ws", status_code="200", quantile="0.9"}/1e6 )) > 5000.0
+ IF          max(sort_desc( scope_appmapper_request_duration_nanoseconds{route!="api_control", route!="api_app_topology_ws", route!="api_app_pipe", route!="api_probe_pipe", status_code="200", quantile="0.9"}/1e6 )) > 5000.0
  FOR         5m
  WITH        { severity="warning" }
  SUMMARY     "app-mapper service: high latency"

monitoring/prometheus/prometheus.yml

@@ -8,8 +8,8 @@ rule_files:
 scrape_configs:
 - job_name: 'kubernetes-services'
   kubernetes_sd_configs:
-  - masters:
-    - 'https://kubernetes.default.svc'
+  - api_servers:
+    - 'https://kubernetes.default.svc.cluster.local'
     in_cluster: true
 
   # https://github.com/prometheus/prometheus/blob/0.16.1/documentation/examples/prometheus-kubernetes.yml