Commit 2c28f8cd by Torkel Ödegaard

Began work on grafana 2.0 install instructions

parent 4244ed33
......@@ -5,184 +5,66 @@ page_keywords: grafana, installation, documentation
---
# Installation
Grafana is easily installed via a Debian/Ubuntu package (.deb), via Redhat/Centos package (.rpm) or manually via
a tar that contains all required files and binaries. If there is not a package or binary for you specific platform you might be able
to build one your self, read [build from source](../project/building_from_source) instructions for more information.
[Download](http://grafana.org/download) the latest release. The release package contain a subfolder, for example **grafana-1.7.0**. The
contents of this folder should be hosted by a web server, for example nginx, apache, IIS. The standard release
packages does not contain a web server to host Grafana.
## From Ubuntu & Debian package
Start by [downloading](http://grafana.org/download/builds) the latest `.deb` package.
### Dependencies
There are no dependencies, Grafana is a client side application that runs in your browser. It only needs a time series store
where it can fetch metrics. If you use InfluxDB Grafana can use it to store dashboards.
If you use Graphite or OpenTSDB you can use Elasticsearch to store dashboards or just use json files stored on disk.
To install the package:
### Provisioning
If you prefer to install grafana via Puppet, Ansible, Docker or Chef. [This page](/docs/provisioning) has compiled a
list of repositories for different provisioning systems
## Configuration
In your chosen Grafana install location, locate the file **config.sample.js** and copy or rename it to **config.js**.
This files contains global settings for your Grafana installation.
### Datasources
The datasources property defines your metric, annotation and dashboard storage backends.
- You can specify multiple datasources.
- default: true marks it as the default metric source (if you have multiple)
- grafanaDB: true marks it for use as dashboard storage (applicable for InfluxDB & Elasticsearch)
### InfluxDB example setup
```javascript
datasources: {
'eu-metrics': {
type: 'influxdb',
url: 'http://my_influxdb_server:8086/db/<db_name>',
username: 'test',
password: 'test',
},
'grafana': {
type: 'influxdb',
url: 'http://my_influxdb_server:8086/db/grafana',
username: 'test',
password: 'test',
grafanaDB: true
},
},
```
In the above example you see two InfluxDB datasources, one for metrics and a seperate used for dashboard storage. You can use the same InfluxDB
database for both. But it is probably a good idea to keep them seperate. The InfluxDB databases need to exist, grafana does not create
them.
### Graphite & Elasticsearch setup example
```javascript
datasources: {
graphite: {
type: 'graphite',
url: "http://my.graphite.server.com:8080",
},
elasticsearch: {
type: 'elasticsearch',
url: "http://my.elastic.server.com:9200",
index: 'grafana-dash',
grafanaDB: true,
}
},
sudo dpkg -i grafana_latest_amd64.deb
```
If you use Graphite you need Elasticsearch if you want to store & search dashboards. You can also use json and scripted dashboards if
you really do not want to setup Elasticsearch.
### OpenTSDB & Elasticsearch setup example
```javascript
datasources: {
opentsdb: {
type: 'opentsdb',
url: "http://my.opentsdb.server.com:4242",
},
elasticsearch: {
type: 'elasticsearch',
url: "http://my.elastic.server.com:9200",
index: 'grafana-dash',
grafanaDB: true,
}
},
```
Please view [this page](/docs/features/opentsdb) for details on how to configure OpenTSDB to work with Grafana.
### Elasticsearch & CORS
As of Elasticsearch v1.4 CORS is disabled by default. This needs to be enabled in the Elasticsearch config file, otherwise grafana will not be able to access Elasticsearch.
## From Redhat & Centos package
Start by [downloading](http://grafana.org/download/builds) the latest `.rpm` package.
```
http.cors.enabled: true
http.cors.allow-origin: *
sudo rpm -Uvh grafana-latest-1.x86_64.rpm
```
Instead of the wildcard you can put your full grafana webserver address (including http:// and port)
### Basic authentication
If your Graphite or Elasticsearch server require basic authentication you can specify the username and password in the url.
For example `"http://admin:secret@my.graphite.com"`
## Global configuration options
```javascript
// specify the limit for dashboard search results
search: {
max_results: 20
},
// default start dashboard
default_route: '/dashboard/file/default.json',
On Redhat/RHEL 5.10 you need to add the grafana user before executing the above.
Execute this to add a grafana user:
// set to false to disable unsaved changes warning
unsaved_changes_warning: true,
```
sudo useradd -r grafana
// set the default timespan for the playlist feature
// Example: "1m", "1h"
playlist_timespan: "1m",
```
// If you want to specify password before saving, please specify it bellow
// The purpose of this password is not security, but to stop some users from accidentally changing dashboards
admin: {
password: ''
},
### Package details
The `.deb` and the `rpm` package install will do the following
// Add your own custom pannels
plugins: {
panels: []
}
```
- Install binaries and frontend files under `/opt/grafana/versions/<version>`
- Symlink dir `/opt/grafana/current` to `/opt/grafana/versions/<version>`
- Symlink `/etc/init.d/grafana` to `/opt/grafana/current/scripts/init.sh`
- Add config file to `/etc/grafana/grafana.ini` , this is where you specify your config settings
- Default configuration is in `/opt/grafana/current/conf/defaults.ini`, do not modify that file
- The default configuration specifies log file at `/var/log/grafana/grafana.log`
- The default configuration specifies sqlite3 db at `/opt/grafana/data/grafana.db`
## Graphite server config
If you haven't used an alternative dashboard for graphite before you need to enable CORS (Cross Origin Resource Sharing).
This is only required if Grafana is hosted on a different web domain from your graphite-web.
### Start the backend & web server
For Apache 2.x:
- Start grafana by `sudo service grafana start`
- This will start the grafana process as the `grafana` user (created during package install)
- Default http port is `3000`, and default user is admin/admin
```javascript
Header set Access-Control-Allow-Origin "*"
Header set Access-Control-Allow-Methods "GET, OPTIONS"
Header set Access-Control-Allow-Headers "origin, authorization, accept"
```
## Manual install from tar file
Start by [downloading](http://grafana.org/download/builds) the latest `.tar.gz` file and extract it.
This will extract into a folder named after the version you downloaded. This folder contains all files required to run grafana.
There are no init scripts or install scripts in this package.
Note that using `"*"` leaves your graphite instance quite open so you might want to consider
using `"http://my.grafana.com"` in place of `"*"`
If your Graphite web is proteced by basic authentication, you have to enable the HTTP verb OPTIONS. Take note that
when using basic auth **Access-Control-Allow-Origin** must not be set to a wildcard, also the header
**Access-Control-Allow-Credentials** must be specified. This looks like the following for Apache:
```html
Header set Access-Control-Allow-Origin "http://mygrafana.com:5656"
Header set Access-Control-Allow-Methods "GET, OPTIONS"
Header set Access-Control-Allow-Headers "origin, authorization, accept"
Header set Access-Control-Allow-Credentials true
<Location />
AuthName "graphs restricted"
AuthType Basic
AuthUserFile /etc/apache2/htpasswd
<LimitExcept OPTIONS>
require valid-user
</LimitExcept>
</Location>
```
To configure grafana add a config file named `custom.ini` to the `conf` folder and override any of the settings defined in
`conf/defaults.ini`. Start grafana by excecuting `./grafana web`. The grafana binary needs the working directory
to be the root install dir (where the binary is and the public folder is located).
For nginx:
## Dependencies
There are no dependencies with the default configuration. You can switch from a sqlite3 database to mysql or postgres but
that is optional. For small to medium setups sqlite3 should suffice.
```javascript
auth_basic "Restricted";
auth_basic_user_file /path/to/my/htpasswd/file;
## Install using provisioning
If you prefer to install grafana via Puppet, Ansible, Docker or Chef. [This page](provisioning) has compiled a
list of repositories for different provisioning systems
if ($http_origin ~* (https?://[^/]*\.somedomain\.com(:[0-9]+)?)) { #Test if request is from allowed domain, you can use multiple if
set $cors "true"; #statements to allow multiple domains, simply setting $cors to true in each one.
}
if ($cors = 'true') {
add_header Access-Control-Allow-Origin $http_origin; #this mirrors back whatever domain the request came from as authorized, as
add_header "Access-Control-Allow-Credentials" "true"; #as long as it matches one of your if statements
add_header "Access-Control-Allow-Methods" "GET, OPTIONS";
add_header "Access-Control-Allow-Headers" "Authorization, origin, accept";
}
```
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment