summaryrefslogtreecommitdiffstats
path: root/packaging/docker/README.md
blob: aec5723e3fb4c565b8325a896deb3f6d0626d730 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
<!--
title: "Install Netdata with Docker"
date: "2020-04-23"
custom_edit_url: "https://github.com/netdata/netdata/edit/master/packaging/docker/README.md"
sidebar_label: "Install Netdata with Docker"
learn_status: "Published"
learn_topic_type: "Tasks"
learn_rel_path: "Installation"
-->

# Install the Netdata Agent with Docker

Running the Netdata Agent in a container works best for an internal network or to quickly analyze a host. Docker helps
you get set up quickly, and doesn't install anything permanent on the system, which makes uninstalling the Agent easy.

See our full list of Docker images at [Docker Hub](https://hub.docker.com/r/netdata/netdata).

Starting with v1.30, Netdata collects anonymous usage information by default and sends it to a self-hosted PostHog instance within the Netdata infrastructure. Read
about the information collected, and learn how to-opt, on our [anonymous statistics](https://github.com/netdata/netdata/blob/master/docs/anonymous-statistics.md)
page.

The usage statistics are _vital_ for us, as we use them to discover bugs and prioritize new features. We thank you for
_actively_ contributing to Netdata's future.

## Limitations running the Agent in Docker

For monitoring the whole host, running the Agent in a container can limit its capabilities. Some data, like the host OS
performance or status, is not accessible or not as detailed in a container as when running the Agent directly on the
host.

A way around this is to provide special mounts to the Docker container so that the Agent can get visibility on host OS
information like `/sys` and `/proc` folders or even `/etc/group` and shadow files.

Also, we now ship Docker images using an [ENTRYPOINT](https://docs.docker.com/engine/reference/builder/#entrypoint)
directive, not a COMMAND directive. Please adapt your execution scripts accordingly. You can find more information about
ENTRYPOINT vs COMMAND in the [Docker
documentation](https://docs.docker.com/engine/reference/builder/#understand-how-cmd-and-entrypoint-interact).

Our POWER8+ Docker images do not support our FreeIPMI collector. This is a technical limitation in FreeIPMI itself,
and unfortunately not something we can realistically work around.

## Create a new Netdata Agent container

> **Notice**: all `docker run` commands and `docker-compose` configurations explicitly set the `nofile` limit. This is
> required on some distros until [14177](https://github.com/netdata/netdata/issues/14177) is resolved. Failure to do so
> may cause a task running in a container to hang and consume 100% of the CPU core.

<details>
<summary>What are these "some distros"?</summary>

If `LimitNOFILE=infinity` results in an open file limit of 1073741816:

```bash
[fedora37 ~]$ docker run --rm busybox grep open /proc/self/limits
Max open files            1073741816           1073741816           files
```
</details>

You can create a new Agent container using either `docker run` or Docker Compose. After using either method, you can
visit the Agent dashboard `http://NODE:19999`.

Both methods create a [bind mount](https://docs.docker.com/storage/bind-mounts/) for Netdata's configuration files
_within the container_ at `/etc/netdata`. See the [configuration section](#configure-agent-containers) for details. If
you want to access the configuration files from your _host_ machine, see [host-editable
configuration](#host-editable-configuration).

**`docker run`**: Use the `docker run` command, along with the following options, to start a new container.

```bash
docker run -d --name=netdata \
  -p 19999:19999 \
  -v netdataconfig:/etc/netdata \
  -v netdatalib:/var/lib/netdata \
  -v netdatacache:/var/cache/netdata \
  -v /etc/passwd:/host/etc/passwd:ro \
  -v /etc/group:/host/etc/group:ro \
  -v /proc:/host/proc:ro \
  -v /sys:/host/sys:ro \
  -v /etc/os-release:/host/etc/os-release:ro \
  --restart unless-stopped \
  --cap-add SYS_PTRACE \
  --security-opt apparmor=unconfined \
  --ulimit nofile=4096 \
  netdata/netdata
```

**Docker Compose**: Copy the following code and paste into a new file called `docker-compose.yml`, then run
`docker-compose up -d` in the same directory as the `docker-compose.yml` file to start the container.

```yaml
version: '3'
services:
  netdata:
    image: netdata/netdata
    container_name: netdata
    hostname: example.com # set to fqdn of host
    ports:
      - 19999:19999
    restart: unless-stopped
    cap_add:
      - SYS_PTRACE
    security_opt:
      - apparmor:unconfined
    ulimits:
      nofile:
        soft: 4096
    volumes:
      - netdataconfig:/etc/netdata
      - netdatalib:/var/lib/netdata
      - netdatacache:/var/cache/netdata
      - /etc/passwd:/host/etc/passwd:ro
      - /etc/group:/host/etc/group:ro
      - /proc:/host/proc:ro
      - /sys:/host/sys:ro
      - /etc/os-release:/host/etc/os-release:ro

volumes:
  netdataconfig:
  netdatalib:
  netdatacache:
```

## Docker tags

The official `netdata/netdata` Docker image provides the following named tags:

* `stable`: The `stable` tag will always point to the most recently published stable build.
* `edge`: The `edge` tag will always point ot the most recently published nightly build. In most cases, this is
  updated daily at around 01:00 UTC.
* `latest`: The `latest` tag will always point to the most recently published build, whether it’s a stable build
  or a nightly build. This is what Docker will use by default if you do not specify a tag.

Additionally, for each stable release, three tags are pushed, one with the full version of the release (for example,
`v1.30.0`), one with just the major and minor version (for example, `v1.30`), and one with just the major version
(for example, `v1`). The tags for the minor versions and major versions are updated whenever a release is published
that would match that tag (for example, if `v1.30.1` were to be published, the `v1.30` tag would be updated to
point to that instead of `v1.30.0`).

## Health Checks

Our Docker image provides integrated support for health checks through the standard Docker interfaces.

You can control how the health checks run by using the environment variable `NETDATA_HEALTHCHECK_TARGET` as follows:

-   If left unset, the health check will attempt to access the
    `/api/v1/info` endpoint of the agent.
-   If set to the exact value 'cli', the health check
    script will use `netdatacli ping` to determine if the agent is running
    correctly or not. This is sufficient to ensure that Netdata did not
    hang during startup, but does not provide a rigorous verification
    that the daemon is collecting data or is otherwise usable.
-   If set to anything else, the health check will treat the value as a
    URL to check for a 200 status code on. In most cases, this should
    start with `http://localhost:19999/` to check the agent running in
    the container.

In most cases, the default behavior of checking the `/api/v1/info`
endpoint will be sufficient. If you are using a configuration which
disables the web server or restricts access to certain APIs, you will
need to use a non-default configuration for health checks to work.

## Configure Agent containers

If you started an Agent container using one of the [recommended methods](#create-a-new-netdata-agent-container), and you
want to edit Netdata's configuration, you must first use `docker exec` to attach to the container. Replace `netdata`
with the name of your container.

```bash
docker exec -it netdata bash
cd /etc/netdata
./edit-config netdata.conf
```

You need to restart the Agent to apply changes. Exit the container if you haven't already, then use the `docker` command
to restart the container: `docker restart netdata`.

### Host-editable configuration

> **Warning**: [edit-config](https://github.com/netdata/netdata/blob/master/docs/configure/nodes.md#the-netdata-config-directory) script doesn't work when executed on
> the host system.

If you want to make your container's configuration directory accessible from the host system, you need to use a
[volume](https://docs.docker.com/storage/bind-mounts/) rather than a bind mount. The following commands create a
temporary `netdata_tmp` container, which is used to populate a `netdataconfig` directory, which is then mounted inside
the container at `/etc/netdata`.

```bash
mkdir netdataconfig
docker run -d --name netdata_tmp netdata/netdata
docker cp netdata_tmp:/etc/netdata netdataconfig/
docker rm -f netdata_tmp
```

**`docker run`**: Use the `docker run` command, along with the following options, to start a new container. Note the
changed `-v $(pwd)/netdataconfig/netdata:/etc/netdata \` line from the recommended example above.

```bash
docker run -d --name=netdata \
  -p 19999:19999 \
  -v $(pwd)/netdataconfig/netdata:/etc/netdata \
  -v netdatalib:/var/lib/netdata \
  -v netdatacache:/var/cache/netdata \
  -v /etc/passwd:/host/etc/passwd:ro \
  -v /etc/group:/host/etc/group:ro \
  -v /proc:/host/proc:ro \
  -v /sys:/host/sys:ro \
  -v /etc/os-release:/host/etc/os-release:ro \
  --restart unless-stopped \
  --cap-add SYS_PTRACE \
  --security-opt apparmor=unconfined \
  --ulimit nofile=4096 \
  netdata/netdata
```

**Docker Compose**: Copy the following code and paste into a new file called `docker-compose.yml`, then run
`docker-compose up -d` in the same directory as the `docker-compose.yml` file to start the container. Note the changed
`./netdataconfig/netdata:/etc/netdata:ro` line from the recommended example above.

```yaml
version: '3'
services:
  netdata:
    image: netdata/netdata
    container_name: netdata
    hostname: example.com # set to fqdn of host
    ports:
      - 19999:19999
    restart: unless-stopped
    cap_add:
      - SYS_PTRACE
    security_opt:
      - apparmor:unconfined
    ulimits:
      nofile:
        soft: 4096
    volumes:
      - ./netdataconfig/netdata:/etc/netdata:ro
      - netdatalib:/var/lib/netdata
      - netdatacache:/var/cache/netdata
      - /etc/passwd:/host/etc/passwd:ro
      - /etc/group:/host/etc/group:ro
      - /proc:/host/proc:ro
      - /sys:/host/sys:ro
      - /etc/os-release:/host/etc/os-release:ro

volumes:
  netdatalib:
  netdatacache:
```

### Change the default hostname

You can change the hostname of a Docker container, and thus the name that appears in the local dashboard and in Netdata
Cloud, when creating a new container. If you want to change the hostname of a Netdata container _after_ you started it,
you can safely stop and remove it. Your configuration and metrics data reside in persistent volumes and are reattached to
the recreated container.

If you use `docker-run`, use the `--hostname` option with `docker run`.

```bash
docker run -d --name=netdata \
  --hostname=my_docker_netdata
```

If you use `docker-compose`, add a `hostname:` key/value pair into your `docker-compose.yml` file, then create the
container again using `docker-compose up -d`.

```yaml
version: '3'
services:
  netdata:
    image: netdata/netdata
    container_name: netdata
    hostname: my_docker_compose_netdata
    ...
```

If you don't want to destroy and recreate your container, you can edit the Agent's `netdata.conf` file directly. See the
above section on [configuring Agent containers](#configure-agent-containers) to find the appropriate method based on
how you created the container.

Alternatively, you can directly use the hostname from the node running the container by mounting
`/etc/hostname` from the host in the container. With `docker run`, this can be done by adding `--volume
/etc/hostname:/etc/hostname:ro` to the options. If you are using Docker Compose, you can add an entry to the
container's `volumes` section reading `- /etc/hostname:/etc/hostname:ro`.

### Add or remove other volumes

Some volumes are optional depending on how you use Netdata:

-   If you don't want to use the apps.plugin functionality, you can remove the mounts of `/etc/passwd` and `/etc/group`
    (they are used to get proper user and group names for the monitored host) to get slightly better security.
-   Most modern linux distros supply `/etc/os-release` although some older distros only supply `/etc/lsb-release`. If
    this is the case you can change the line above that mounts the file inside the container to
    `-v /etc/lsb-release:/host/etc/lsb-release:ro`.
-   If your host is virtualized then Netdata cannot detect it from inside the container and will output the wrong
    metadata (e.g. on `/api/v1/info` queries). You can fix this by setting a variable that overrides the detection
    using, e.g. `--env VIRTUALIZATION=$(systemd-detect-virt -v)`. If you are using a `docker-compose.yml` then add:

```yaml
    environment:
      - VIRTUALIZATION=${VIRTUALIZATION}
```

This allows the information to be passed into `docker-compose` using:

```bash
VIRTUALIZATION=$(systemd-detect-virt -v) docker-compose up
```

#### Files inside systemd volumes

If a volume is used by systemd service, some files can be removed during 
[reinitialization](https://github.com/netdata/netdata/issues/9916). To avoid this, you need to add
`RuntimeDirectoryPreserve=yes` to the service file.

### Docker container names resolution

There are a few options for resolving container names within Netdata. Some methods of doing so will allow root access to
your machine from within the container. Please read the following carefully.

#### Docker socket proxy (safest option)

Deploy a Docker socket proxy that accepts and filters out requests using something like
[HAProxy](https://github.com/netdata/netdata/blob/master/docs/Running-behind-haproxy.md) so that it restricts connections to read-only access to the CONTAINERS
endpoint.

The reason it's safer to expose the socket to the proxy is because Netdata has a TCP port exposed outside the Docker
network. Access to the proxy container is limited to only within the network.

Below is [an example repository (and image)](https://github.com/Tecnativa/docker-socket-proxy) that provides a proxy to
the socket.

You run the Docker Socket Proxy in its own Docker Compose file and leave it on a private network that you can add to
other services that require access.

```yaml
version: '3'
services:
  netdata:
    image: netdata/netdata
    # ... rest of your config ...
    ports:
      - 19999:19999
    environment:
      - DOCKER_HOST=proxy:2375
  proxy:
    image: tecnativa/docker-socket-proxy
    volumes:
     - /var/run/docker.sock:/var/run/docker.sock:ro
    environment:
      - CONTAINERS=1

```
**Note:** Replace `2375` with the port of your proxy.

#### Giving group access to the Docker socket (less safe)

> You should seriously consider the necessity of activating this option, as it grants to the `netdata`
user access to the privileged socket connection of docker service and therefore your whole machine.

If you want to have your container names resolved by Netdata, make the `netdata` user be part of the group that owns the
socket.

```yaml
version: '3'
services:
  netdata:
    image: netdata/netdata
    # ... rest of your config ...
    volumes:
      # ... other volumes ...
      - /var/run/docker.sock:/var/run/docker.sock:ro
    environment:
      - PGID=[GROUP NUMBER]
```

To achieve that just add environment variable `PGID=[GROUP NUMBER]` to the Netdata container, where `[GROUP NUMBER]` is
practically the group id of the group assigned to the docker socket, on your host.

This group number can be found by running the following (if socket group ownership is docker):

```bash
grep docker /etc/group | cut -d ':' -f 3
```

#### Running as root (unsafe)

> You should seriously consider the necessity of activating this option, as it grants to the `netdata` user access to
> the privileged socket connection of docker service, and therefore your whole machine.

```yaml
version: '3'
services:
  netdata:
    image: netdata/netdata
    # ... rest of your config ...
    volumes:
      # ... other volumes ...
      - /var/run/docker.sock:/var/run/docker.sock:ro
    environment:
      - DOCKER_USR=root
```

### Docker container network interfaces monitoring

Netdata can map a virtual interface in the system namespace to an interface inside a Docker container
when using network [bridge](https://docs.docker.com/network/bridge/) driver. To do this, the Netdata container needs
additional privileges:

- the host PID mode. This turns on sharing between container and the host operating system the PID
  address space (needed to get list of PIDs from `cgroup.procs` file).

- `SYS_ADMIN` capability (needed to execute `setns()`).

**docker run**:

```bash
docker run -d --name=netdata \
  ...
  --pid=host \
  --cap-add SYS_ADMIN \
  ...
  netdata/netdata
```

**docker compose**:

```yaml
version: '3'
services:
  netdata:
    image: netdata/netdata
    container_name: netdata
    pid: host
    cap_add:
      - SYS_ADMIN
    ...
```

### Pass command line options to Netdata

Since we use an [ENTRYPOINT](https://docs.docker.com/engine/reference/builder/#entrypoint) directive, you can provide
[Netdata daemon command line options](https://github.com/netdata/netdata/blob/master/daemon/README.md#command-line-options) such as the IP address Netdata will be
running on, using the [command instruction](https://docs.docker.com/engine/reference/builder/#cmd). 

## Install the Agent using Docker Compose with SSL/TLS enabled HTTP Proxy

For a permanent installation on a public server, you should [secure the Netdata
instance](https://github.com/netdata/netdata/blob/master/docs/netdata-security.md). This section contains an example of how to install Netdata with an SSL
reverse proxy and basic authentication.

You can use the following `docker-compose.yml` and Caddyfile files to run Netdata with Docker. Replace the domains and
email address for [Let's Encrypt](https://letsencrypt.org/) before starting.

### Caddyfile

This file needs to be placed in `/opt` with name `Caddyfile`. Here you customize your domain, and you need to provide
your email address to obtain a Let's Encrypt certificate. Certificate renewal will happen automatically and will be
executed internally by the caddy server.

```caddyfile
netdata.example.org {
  reverse_proxy netdata:19999
  tls admin@example.org
}
```

### docker-compose.yml

After setting Caddyfile run this with `docker-compose up -d` to have fully functioning Netdata setup behind HTTP reverse
proxy.

```yaml
version: '3'
volumes:
  caddy_data:
  caddy_config:

services:
  caddy:
    image: caddy:2
    ports:
      - "80:80"
      - "443:443"
    volumes:
      - /opt/Caddyfile:/etc/caddy/Caddyfile
      - caddy_data:/data
      - caddy_config:/config
  netdata:
    restart: always
    hostname: netdata.example.org
    image: netdata/netdata
    cap_add:
      - SYS_PTRACE
    security_opt:
      - apparmor:unconfined
    ulimits:
      nofile:
        soft: 4096
    volumes:
      - netdatalib:/var/lib/netdata
      - netdatacache:/var/cache/netdata
      - /etc/passwd:/host/etc/passwd:ro
      - /etc/group:/host/etc/group:ro
      - /proc:/host/proc:ro
      - /sys:/host/sys:ro
      - /var/run/docker.sock:/var/run/docker.sock:ro

volumes:
  netdatalib:
  netdatacache:
```

### Restrict access with basic auth

You can restrict access by following [official caddy guide](https://caddyserver.com/docs/caddyfile/directives/basicauth#basicauth) and adding lines to
Caddyfile.

## Publish a test image to your own repository

At Netdata, we provide multiple ways of testing your Docker images using your own repositories.
You may either use the command line tools available or take advantage of our GitHub Acions infrastructure.