master
md 258 lines 14.5 KB
Rendered Raw
1 # Netdata, Prometheus, Grafana stack
2
3 ## Intro
4
5 In this article I will walk you through the basics of getting Netdata, Prometheus and Grafana all working together and
6 monitoring your application servers. This article will be using docker on your local workstation. We will be working
7 with docker in an ad-hoc way, launching containers that run `/bin/bash` and attaching a TTY to them. I use docker here
8 in a purely academic fashion and do not condone running Netdata in a container. I pick this method so individuals
9 without cloud accounts or access to VMs can try this out and for it's speed of deployment.
10
11 ## Why Netdata, Prometheus, and Grafana
12
13 Some time ago I was introduced to Netdata by a coworker. We were attempting to troubleshoot python code which seemed to
14 be bottlenecked. I was instantly impressed by the amount of metrics Netdata exposes to you. I quickly added Netdata to
15 my set of go-to tools when troubleshooting systems performance.
16
17 Some time ago, even later, I was introduced to Prometheus. Prometheus is a monitoring application which flips the normal
18 architecture around and polls rest endpoints for its metrics. This architectural change greatly simplifies and decreases
19 the time necessary to begin monitoring your applications. Compared to current monitoring solutions the time spent on
20 designing the infrastructure is greatly reduced. Running a single Prometheus server per application becomes feasible
21 with the help of Grafana.
22
23 Grafana has been the go to graphing tool for… some time now. It's awesome, anyone that has used it knows it's awesome.
24 We can point Grafana at Prometheus and use Prometheus as a data source. This allows a pretty simple overall monitoring
25 architecture: Install Netdata on your application servers, point Prometheus at Netdata, and then point Grafana at
26 Prometheus.
27
28 I'm omitting an important ingredient in this stack in order to keep this tutorial simple and that is service discovery.
29 My personal preference is to use Consul. Prometheus can plug into consul and automatically begin to scrape new hosts
30 that register a Netdata client with Consul.
31
32 At the end of this tutorial you will understand how each technology fits together to create a modern monitoring stack.
33 This stack will offer you visibility into your application and systems performance.
34
35 ## Getting Started - Netdata
36
37 To begin let's create our container which we will install Netdata on. We need to run a container, forward the necessary
38 port that Netdata listens on, and attach a tty so we can interact with the bash shell on the container. But before we do
39 this we want name resolution between the two containers to work. In order to accomplish this we will create a
40 user-defined network and attach both containers to this network. The first command we should run is:
41
42 ```sh
43 docker network create --driver bridge netdata-tutorial
44 ```
45
46 With this user-defined network created we can now launch our container we will install Netdata on and point it to this
47 network.
48
49 ```sh
50 docker run -it --name netdata --hostname netdata --network=netdata-tutorial -p 19999:19999 centos:latest '/bin/bash'
51 ```
52
53 This command creates an interactive tty session (`-it`), gives the container both a name in relation to the docker
54 daemon and a hostname (this is so you know what container is which when working in the shells and docker maps hostname
55 resolution to this container), forwards the local port 19999 to the container's port 19999 (`-p 19999:19999`), sets the
56 command to run (`/bin/bash`) and then chooses the base container images (`centos:latest`). After running this you should
57 be sitting inside the shell of the container.
58
59 After we have entered the shell we can install Netdata. This process could not be easier. If you take a look at [this
60 link](/packaging/installer/README.md), the Netdata devs give us several one-liners to install Netdata. I have not had
61 any issues with these one liners and their bootstrapping scripts so far (If you guys run into anything do share). Run
62 the following command in your container.
63
64 <!-- candidate for reuse -->
65 ```sh
66 wget -O /tmp/netdata-kickstart.sh https://get.netdata.cloud/kickstart.sh && sh /tmp/netdata-kickstart.sh --dont-wait
67 ```
68
69 After the install completes you should be able to hit the Netdata dashboard at <http://localhost:19999/> (replace
70 localhost if you're doing this on a VM or have the docker container hosted on a machine not on your local system). If
71 this is your first time using Netdata I suggest you take a look around. The amount of time I've spent digging through
72 `/proc` and calculating my own metrics has been greatly reduced by this tool. Take it all in.
73
74 Next I want to draw your attention to a particular endpoint. Navigate to
75 <http://localhost:19999/api/v1/allmetrics?format=prometheus&help=yes> In your browser. This is the endpoint which
76 publishes all the metrics in a format which Prometheus understands. Let's take a look at one of these metrics.
77 `netdata_disk_space_GiB_average{chart="disk_space._run",dimension="avail",family="/run",mount_point="/run",filesystem="tmpfs",mount_root="/"} 0.0298195 1684951093000`
78 This metric is representing several things which I will go in more details in the section on Prometheus. For now understand
79 that this metric: `netdata_disk_space_GiB_average` has several labels: (`chart`, `family`, `dimension`, `mountt_point`, `filesystem`, `mount_root`).
80 This corresponds with disk space you see on the Netdata dashboard.
81
82 ![](https://github.com/ldelossa/NetdataTutorial/raw/master/Screen%20Shot%202017-07-28%20at%204.00.45%20PM.png)
83
84 This CHART is called `system.cpu`, The FAMILY is `cpu`, and the DIMENSION we are observing is `system`. You can begin to
85 draw links between the charts in Netdata to the Prometheus metrics format in this manner.
86
87 ## Prometheus
88
89 We will be installing Prometheus in a container for purpose of demonstration. While Prometheus does have an official
90 container I would like to walk through the install process and setup on a fresh container. This will allow anyone
91 reading to migrate this tutorial to a VM or Server of any sort.
92
93 Let's start another container in the same fashion as we did the Netdata container.
94
95 ```sh
96 docker run -it --name prometheus --hostname prometheus \
97 --network=netdata-tutorial -p 9090:9090 centos:latest '/bin/bash'
98 ```
99
100 This should drop you into a shell once again. Once there quickly install your favorite editor as we will be editing
101 files later in this tutorial.
102
103 ```sh
104 yum install vim -y
105 ```
106
107 You will also need `wget` and `curl` to download files and `sudo` if you are not root.
108
109 ```sh
110 yum install curl sudo wget -y
111 ```
112
113 Prometheus provides a tarball of their latest stable versions [here](https://prometheus.io/download/).
114
115 Let's download the latest version and install into your container.
116
117 ```sh
118 cd /tmp && curl -s https://api.github.com/repos/prometheus/prometheus/releases/latest \
119 | grep "browser_download_url.*linux-amd64.tar.gz" \
120 | cut -d '"' -f 4 \
121 | wget -qi -
122
123 mkdir /opt/prometheus
124
125 sudo tar -xvf /tmp/prometheus-*linux-amd64.tar.gz -C /opt/prometheus --strip=1
126 ```
127
128 This should get Prometheus installed into the container. Let's test that we can run Prometheus and connect to it's web
129 interface.
130
131 ```sh
132 /opt/prometheus/prometheus --config.file=/opt/prometheus/prometheus.yml
133 ```
134
135 Now attempt to go to <http://localhost:9090/>. You should be presented with the Prometheus homepage. This is a good
136 point to talk about Prometheus's data model which can be viewed here: <https://prometheus.io/docs/concepts/data_model/>
137 As explained we have two key elements in Prometheus metrics. We have the _metric_ and its _labels_. Labels allow for
138 granularity between metrics. Let's use our previous example to further explain.
139
140 ```text
141 netdata_disk_space_GiB_average{chart="disk_space._run",dimension="avail",family="/run",mount_point="/run",filesystem="tmpfs",mount_root="/"} 0.0298195 1684951093000
142 ```
143
144 Here our metric is `netdata_disk_space_GiB_average` and our common labels are `chart`, `family`, and `dimension`. The
145 last two values constitute the actual metric value for the metric type (gauge, counter, etc…). We also have specific
146 label for this chart named `mount_point`,`filesystem`, and `mount_root`. We can begin graphing system metrics with this information,
147 but first we need to hook up Prometheus to poll Netdata stats.
148
149 Let's move our attention to Prometheus's configuration. Prometheus gets it config from the file located (in our example)
150 at `/opt/prometheus/prometheus.yml`. I won't spend an extensive amount of time going over the configuration values
151 documented here: <https://prometheus.io/docs/operating/configuration/>. We will be adding a new job under the
152 `scrape_configs`. Let's make the `scrape_configs` section look like this (we can use the DNS name Netdata due to the
153 custom user-defined network we created in docker beforehand).
154
155 ```yaml
156 scrape_configs:
157 # The job name is added as a label `job=<job_name>` to any timeseries scraped from this config.
158 - job_name: 'prometheus'
159
160 # metrics_path defaults to '/metrics'
161 # scheme defaults to 'http'.
162
163 static_configs:
164 - targets: ['localhost:9090']
165
166 - job_name: 'netdata'
167
168 metrics_path: /api/v1/allmetrics
169 params:
170 format: [ prometheus ]
171
172 static_configs:
173 - targets: ['netdata:19999']
174 ```
175
176 Let's start Prometheus once again by running `/opt/prometheus/prometheus`. If we now navigate to Prometheus at
177 <http://localhost:9090/targets> we should see our target being successfully scraped. If we now go back to the
178 Prometheus's homepage and begin to type `netdata\_` Prometheus should auto complete metrics it is now scraping.
179
180 ![](https://github.com/ldelossa/NetdataTutorial/raw/master/Screen%20Shot%202017-07-28%20at%205.13.43%20PM.png)
181
182 Let's now start exploring how we can graph some metrics. Back in our Netdata container lets get the CPU spinning with a
183 pointless busy loop. On the shell do the following:
184
185 ```sh
186 [root@netdata /]# while true; do echo "HOT HOT HOT CPU"; done
187 ```
188
189 Our Netdata cpu graph should be showing some activity. Let's represent this in Prometheus. In order to do this let's
190 keep our metrics page open for reference: <http://localhost:19999/api/v1/allmetrics?format=prometheus&help=yes>. We are
191 setting out to graph the data in the CPU chart so let's search for `system.cpu` in the metrics page above. We come
192 across a section of metrics with the first comments `# COMMENT homogeneous chart "system.cpu", context "system.cpu",
193 family "cpu", units "percentage"` followed by the metrics. This is a good start now let us drill down to the specific
194 metric we would like to graph.
195
196 ```text
197 # COMMENT
198 netdata_system_cpu_percentage_average: dimension "system", value is percentage, gauge, dt 1501275951 to 1501275951 inclusive
199 netdata_system_cpu_percentage_average{chart="system.cpu",family="cpu",dimension="system"} 0.0000000 1501275951000
200 ```
201
202 Here we learn that the metric name we care about is `netdata_system_cpu_percentage_average` so throw this into
203 Prometheus and see what we get. We should see something similar to this (I shut off my busy loop)
204
205 ![](https://github.com/ldelossa/NetdataTutorial/raw/master/Screen%20Shot%202017-07-28%20at%205.47.53%20PM.png)
206
207 This is a good step toward what we want. Also make note that Prometheus will tag on an `instance` label for us which
208 corresponds to our statically defined job in the configuration file. This allows us to tailor our queries to specific
209 instances. Now we need to isolate the dimension we want in our query. To do this let us refine the query slightly. Let's
210 query the dimension also. Place this into our query text box.
211 `netdata_system_cpu_percentage_average{dimension="system"}` We now wind up with the following graph.
212
213 ![](https://github.com/ldelossa/NetdataTutorial/raw/master/Screen%20Shot%202017-07-28%20at%205.54.40%20PM.png)
214
215 Awesome, this is exactly what we wanted. If you haven't caught on yet we can emulate entire charts from Netdata by using
216 the `chart` dimension. If you'd like you can combine the `chart` and `instance` dimension to create per-instance charts.
217 Let's give this a try: `netdata_system_cpu_percentage_average{chart="system.cpu", instance="netdata:19999"}`
218
219 This is the basics of using Prometheus to query Netdata. I'd advise everyone at this point to read [this
220 page](/src/exporting/prometheus/README.md#using-netdata-with-prometheus). The key point here is that Netdata can export metrics from
221 its internal DB or can send metrics _as-collected_ by specifying the `source=as-collected` URL parameter like so.
222 <http://localhost:19999/api/v1/allmetrics?format=prometheus&help=yes&types=yes&source=as-collected> If you choose to use
223 this method you will need to use Prometheus's set of functions here: <https://prometheus.io/docs/querying/functions/> to
224 obtain useful metrics as you are now dealing with raw counters from the system. For example you will have to use the
225 `irate()` function over a counter to get that metric's rate per second. If your graphing needs are met by using the
226 metrics returned by Netdata's internal database (not specifying any source= URL parameter) then use that. If you find
227 limitations then consider re-writing your queries using the raw data and using Prometheus functions to get the desired
228 chart.
229
230 ## Grafana
231
232 Finally we make it to grafana. This is the easiest part in my opinion. This time we will actually run the official
233 grafana docker container as all configuration we need to do is done via the GUI. Let's run the following command:
234
235 ```sh
236 docker run -i -p 3000:3000 --network=netdata-tutorial grafana/grafana
237 ```
238
239 This will get grafana running at <http://localhost:3000/>. Let's go there and
240 login using the credentials Admin:Admin.
241
242 The first thing we want to do is click "Add data source". Let's make it look like the following screenshot
243
244 ![](https://github.com/ldelossa/NetdataTutorial/raw/master/Screen%20Shot%202017-07-28%20at%206.36.55%20PM.png)
245
246 With this completed let's graph! Create a new Dashboard by clicking on the top left Grafana Icon and create a new graph
247 in that dashboard. Fill in the query like we did above and save.
248
249 ![](https://github.com/ldelossa/NetdataTutorial/raw/master/Screen%20Shot%202017-07-28%20at%206.39.38%20PM.png)
250
251 ## Conclusion
252
253 There you have it, a complete systems monitoring stack which is very easy to deploy. From here I would begin to
254 understand how Prometheus and a service discovery mechanism such as Consul can play together nicely. My current prod
255 deployments automatically register Netdata services into Consul and Prometheus automatically begins to scrape them. Once
256 achieved you do not have to think about the monitoring system until Prometheus cannot keep up with your scale. Once this
257 happens there are options presented in the Prometheus documentation for solving this. Hope this was helpful, happy
258 monitoring.