Merge pull request #29 from thomasjpfan/issue-10

Adds Flexible Labeling

Dockerfile

@@ -3,7 +3,7 @@ ADD . /src
 WORKDIR /src
 RUN go get -t github.com/stretchr/testify/suite
 RUN go get -d -v -t
-RUN go test --cover ./... --run UnitTest
+RUN go test --cover ./... --run UnitTest -p 1
 RUN CGO_ENABLED=0 GOOS=linux go build -v -o docker-flow-monitor
 
 

docs/config.md

@@ -103,8 +103,13 @@ curl `[IP_OF_ONE_OF_SWARM_NODES]:8080/v1/docker-flow-monitor/reconfigure?scrapeP
 
 Please consult [Prometheus Configuration](https://prometheus.io/docs/operating/configuration/) for more information about the available options.
 
-## Scrapes
+## Scrape Secret Configuration
 
 Additional scrapes can be added through files prefixed with `scrape_`. By default, all such files located in `/run/secrets` are automatically added to the `scrape_configs` section of the configuration. The directory can be changed by setting a different value to the environment variable `CONFIGS_DIR`.
 
 The simplest way to add scrape configs is to use Docker [secrets](https://docs.docker.com/engine/swarm/secrets/) or [configs](https://docs.docker.com/engine/swarm/configs/).
+
+
+## Scrape Label Configuration
+
+When using a version of [Docker Flow Swarm Listener](https://github.com/vfarcic/docker-flow-swarm-listener), DFSL, newer than `18.02.06-31`, you can configure DFSL to send node node hostnames to `Docker Flow Monitor`, DFM. This can be done by setting `DF_INCLUDE_NODE_IP_INFO` to `true` in the DFSL environment. DFM will automatically display the node hostnames as a label for each prometheus target. The `DF_SCRAPE_TARGET_LABELS` env variable allows for additional labels to be displayed. For example, if a service has env variables `com.df.env=prod` and `com.df.domain=frontend`, you can set `DF_SCRAPE_TARGET_LABELS=env,domain` in DFM to display the `prod` and `frontend` labels in prometheus.

docs/tutorial-flexible-labeling.md

@@ -0,0 +1,110 @@
+# Flexible Labeling with Docker Flow Monitor
+
+*Docker Flow Monitor* and *Docker Flow Swarm Listener* can be configured to allow for more flexible labeling of exporters. Please read the  [Running Docker Flow Monitor](tutorial.md) tutorial before reading this one. This tutorial focuses on configuring the stacks to allow for flexible labeling.
+
+## Setting Up A Cluster
+
+!!! info
+    Feel free to skip this section if you already have a Swarm cluster that can be used for this tutorial
+
+We'll create a Swarm cluster consisting of three nodes created with Docker Machine.
+
+```bash
+git clone https://github.com/vfarcic/docker-flow-monitor.git
+
+cd docker-flow-monitor
+
+./scripts/dm-swarm.sh
+
+eval $(docker-machine env swarm-1)
+```
+
+## Deploying Docker Flow Monitor
+
+We will deploy [stacks/docker-flow-monitor-flexible-labels.yml](https://github.com/vfarcic/docker-flow-monitor/blob/master/stacks/docker-flow-monitor-flexible-labels.yml) stack that contains three services: `monitor`, `alert-manager` and `swarm-listener`. The `swarm-listener` service includes an additional environment variable: `DF_INCLUDE_NODE_IP_INFO=true`. This configures `swarm-listener` to send node and ip information to `mointor`.
+
+The `monitor` service includes the environment variable: `DF_SCRAPE_TARGET_LABELS=env,metricType`. This sets up flexible labeling for exporters. If an exporter defines a deploy label `com.df.env` or `com.df.metricType`, that label will be used by `monitor`.
+
+Let's deploy the `monitor` stack:
+
+```bash
+docker network create -d overlay monitor
+
+docker stack deploy \
+    -c stacks/docker-flow-monitor-flexible-labels.yml \
+    monitor
+```
+
+## Collecting Metrics and Defining Alerts
+
+We will deploy exporters stack defined in [stacks/exporters-tutorial-flexible-labels.yml](https://github.com/vfarcic/docker-flow-monitor/blob/master/stacks/exporters-tutorial-flexible-labels.yml),  two containing two services: `cadvisor` and `node-exporter`.
+
+The definition of the `cadvisor` service contains additional deploy labels:
+
+```yaml
+  cadvisor:
+    image: google/cadvisor
+    networks:
+      - monitor
+    ...
+    deploy:
+      mode: global
+      labels:
+        ...
+        - com.df.scrapeNetwork=monitor
+        - com.df.env=prod
+        - com.df.metricType=system
+```
+
+The `com.df.scrapeNetwork` deploy label tells `swarm-listener` to use `cadvisor`'s IP on the `monitor` network. This is important because the `monitor` service is using the `monitor` network to scrape `cadvisor`. The `com.df.env=prod` and `com.df.metricType=system` deploy labels configures flexible labeling for `cadvisor`.
+
+The second service, `node-exporter` is also configured with flexiable labels:
+
+```yaml
+  node-exporter:
+    image: basi/node-exporter
+    networks:
+      - monitor
+    ...
+    deploy:
+      mode: global
+      labels:
+        ...
+        - com.df.scrapeNetwork=monitor
+        - com.df.env=dev
+        - com.df.metricType=system
+```
+
+Let's deploy the `exporter` stack
+
+```bash
+docker stack deploy \
+    -c stacks/exporters-tutorial-flexible-labels.yml \
+    exporter
+```
+
+Please wait until the service in the stack are up-and-running. You can check their status by executing `docker stack ps exporter`.
+
+Now we can open the *Prometheus* targets page from a browser.
+
+> If you're a Windows user, Git Bash might not be able to use the `open` command. If that's the case, replace the `open` command with `echo`. As a result, you'll get the full address that should be opened directly in your browser of choice.
+
+```bash
+open "http://$(docker-machine ip swarm-1):9090/targets"
+```
+
+You should see a targets page similar to the following:
+
+![Flexiable Labeling Targets Page](img/flexiable-labeling-targets-page.png)
+
+Each service is labeled with its associated `com.df.env` or `com.df.metricType` deploy label. In addition, the `node` label is the hostname the service is running on.
+
+## What Now?
+
+*Docker Flow Monitors*'s flexible labeling feature provides more information about your services. Please consult the documentation for any additional information you might need. Feel free to open [an issue](https://github.com/vfarcic/docker-flow-monitor/issues) if you require additional info, if you find a bug, or if you have a feature request.
+
+Before you go, please remove the cluster we created and free those resources for something else.
+
+```bash
+docker-machine rm -f swarm-1 swarm-2 swarm-3
+```

mkdocs.yml

@@ -4,6 +4,7 @@ pages:
   - Tutorial:
     - Running Docker Flow Monitor: tutorial.md
     - Auto-Scaling Services Using Instrumented Metrics: auto-scaling.md
+    - Flexible Labeling with Docker Flow Monitor: tutorial-flexible-labeling.md
   - Configuration: config.md
   - Usage: usage.md
   - Migration Guide: migration.md

prometheus/config.go

@@ -2,6 +2,7 @@ package prometheus
 
 import (
 	"bytes"
+	"encoding/json"
 	"fmt"
 	"net/url"
 	"os"
@@ -18,14 +19,17 @@ import (
 // WriteConfig creates Prometheus configuration at configPath and writes alerts into /etc/prometheus/alert.rules
 func WriteConfig(configPath string, scrapes map[string]Scrape, alerts map[string]Alert) {
 	c := &Config{}
+	fileSDDir := "/etc/prometheus/file_sd"
+	alertRulesPath := "/etc/prometheus/alert.rules"
 
 	configDir := filepath.Dir(configPath)
 	FS.MkdirAll(configDir, 0755)
+	FS.MkdirAll(fileSDDir, 0755)
 	c.InsertScrapes(scrapes)
 
 	if len(alerts) > 0 {
 		logPrintf("Writing to alert.rules")
-		afero.WriteFile(FS, "/etc/prometheus/alert.rules", []byte(GetAlertConfig(alerts)), 0644)
+		afero.WriteFile(FS, alertRulesPath, []byte(GetAlertConfig(alerts)), 0644)
 		c.RuleFiles = []string{"alert.rules"}
 	}
 
@@ -35,6 +39,7 @@ func WriteConfig(configPath string, scrapes map[string]Scrape, alerts map[string
 			logPrintf("Unable to insert alertmanager url %s into prometheus config", alertmanagerURL)
 		}
 	}
+	c.CreateFileStaticConfig(scrapes, fileSDDir)
 
 	for _, e := range os.Environ() {
 		envSplit := strings.SplitN(e, "=", 2)
@@ -98,6 +103,9 @@ func (c *Config) InsertScrapes(scrapes map[string]Scrape) {
 		if len(metricsPath) == 0 {
 			metricsPath = "/metrics"
 		}
+		if s.NodeInfo != nil && len(*s.NodeInfo) > 0 {
+			continue
+		}
 		if s.ScrapeType == "static_configs" {
 			newScrape = &ScrapeConfig{
 				ServiceDiscoveryConfig: ServiceDiscoveryConfig{
@@ -152,6 +160,63 @@ func (c *Config) InsertScrapesFromDir(dir string) {
 
 }
 
+// CreateFileStaticConfig creates static config files
+func (c *Config) CreateFileStaticConfig(scrapes map[string]Scrape, fileSDDir string) {
+
+	staticFiles := map[string]struct{}{}
+	for _, s := range scrapes {
+		fsc := FileStaticConfig{}
+		if s.NodeInfo == nil {
+			continue
+		}
+		for n := range *s.NodeInfo {
+			tg := TargetGroup{}
+			tg.Targets = []string{fmt.Sprintf("%s:%d", n.Addr, s.ScrapePort)}
+			tg.Labels = map[string]string{}
+			if s.ScrapeLabels != nil {
+				for k, v := range *s.ScrapeLabels {
+					tg.Labels[k] = v
+				}
+			}
+			tg.Labels["node"] = n.Name
+			tg.Labels["service"] = s.ServiceName
+			fsc = append(fsc, &tg)
+		}
+
+		if len(fsc) == 0 {
+			continue
+		}
+
+		fscBytes, err := json.Marshal(fsc)
+		if err != nil {
+			continue
+		}
+		filePath := fmt.Sprintf("%s/%s.json", fileSDDir, s.ServiceName)
+		afero.WriteFile(FS, filePath, fscBytes, 0644)
+		newScrape := &ScrapeConfig{
+			ServiceDiscoveryConfig: ServiceDiscoveryConfig{
+				FileSDConfigs: []*SDConfig{{
+					Files: []string{filePath},
+				}},
+			},
+			JobName: s.ServiceName,
+		}
+		c.ScrapeConfigs = append(c.ScrapeConfigs, newScrape)
+		staticFiles[filePath] = struct{}{}
+	}
+
+	// Remove scrapes that are not in fileStaticServices
+	currentStaticFiles, err := afero.Glob(FS, fmt.Sprintf("%s/*.json", fileSDDir))
+	if err != nil {
+		return
+	}
+	for _, file := range currentStaticFiles {
+		if _, ok := staticFiles[file]; !ok {
+			FS.Remove(file)
+		}
+	}
+}
+
 func normalizeScrapeFile(content []byte) []byte {
 	spaceCnt := 0
 	for i, c := range content {