Nginx Nginx - Perfect install for netdata (system monitor) - Step by step

I am just getting in to netdata, so can't be of much help at the moment, becides I am a regular noob. My end goal is to have netdata to monitor not only the hardware, but also nginx (I see that you have that already), php-fpm, mariadb, and redis.

I will post what I found here if you don't mind.

 

Adding a php-fpm monitoring is very much similar to nginx one - you need to enable the status page and then give this url to netdata for monitoring.

Steps:

1. Enable the php status page. On a default centminmod install you just need to uncomment:

Code:

#include /usr/local/nginx/conf/phpstatus.conf;

from /usr/local/nginx/conf/conf.d/virtual.conf

Then restart nginx and php-fpm.

Full instructions here in the section How To Enable Php-Fpm Usage Stats Display ? : PHP-FPM - CentminMod.com LEMP Nginx web stack for CentOS

2. Create (if it doesn't exist already) a file phpfpm.conf in /etc/netdata/python.d with the following contents:

Code:

local:
  url     : 'http://localhost/phpstatus'

3. Restart netdata:

Code:

service netdata restart

Thats it! Now you should be able to see php-fpm stats in the right menu under PHP-FPM Local.

 

Succeeded with mariadb as well, but it was more tricky.

1. Create a user for netdata to use to access the database.

Code:

# mysql -u root -p
MariaDB [(none)]> CREATE USER 'netdata'@'localhost';
MariaDB [(none)]> GRANT USAGE on *.* to 'netdata'@'localhost';
MariaDB [(none)]> FLUSH PRIVILEGES;
MariaDB [(none)]> exit

Commands taken from: How to Monitor MySQL/MariaDB Databases using Netdata on CentOS 7

2. Install the required python libraries as outlined at the top of the page here: mysql - Netdata Documentation In my case it was simply running:

Code:

pip install mysqlclient

3. Create mysql.conf in /etc/netdata/python.d/ with the following contents:

Code:

local:
  'my.cnf'   : '/etc/my.cnf'
  user     : 'netdata'

I am not even sure that the first line is needed too.

4. Restart netdata and enjoy the MySQL Local stats in the right menu.

I will explore at a later point whether I need this: User Statistics For now, it looks sufficient.

 

Redis is also done.

If you are having redis server listening to the default port, all it takes is creating redis.conf in /etc/netdata/python.d/ with the following contents:

Code:

socket:
  name     : 'local'
  socket   : '/var/lib/redis/redis.sock'

localhost:
  name     : 'local'
  host     : 'localhost'
  port     : 6379

Info taken from the docs: redis - Netdata Documentation

 

Looking forward to that part, at least some pointers would be helpful. I am not able to find global.ini at all.

Edit: I added in the [web] part of netdata.conf the following string:

Code:

allow connections from = localhost <my ip>

Seems to work fine. Per this comment from the docs, it seems like the most restrictive option. Good enough for now for me.

 

I managed to setup the email alerting as well. As I don't intend to maintain a mail server I am using Amazon SES. Here are the settings to setup Amazon SES with sendmail for linux:

Integrating Amazon SES with Sendmail - Amazon Simple Email Service

I suggest to follow this guide to the letter, because even though it is not hard, there are some fine details. Also keep in mind that sending the test email takes a few minutes after you send it with ctrl+d, I was impatient that I didn't do it correctly In case of issues check:

Code:

/var/mail/<username>

Most probably the exact error will be logged there.

After sendmail is properly setup, you need to setup netdata to send emails as well. Instructions pointing out which files you need to edit for from, to addresses, severity and sending test alert: email - Netdata Documentation

There are many settings to tweak, I figured I'll try the defaults first and depending on the spam level and my personal feeling how many of them are false alarms I will tweak further.

Thats it I feel I am happy with my setup now, my main target is to monitor the soft raid for failed drives, so everything else is extra.

 

Of course no issues, especially if you want to create an easy to follow complete guide.

It will be useful for me down the road as well, I can imagine that if I have to reinstall everything in three years a well laid out guide will be very useful for all the steps I am skipping right now in my description, but are probably important.

 

Another fairly easy to enable plugin is elastic search. You only need elasticsearch.conf in /etc/netdata/python.d with the following contents:

Code (Text):

local:
  host               : 'localhost'    # Elasticsearch server ip address or hostname
  port               : '9200'         # Port on which elasticsearch listens

localhost and 9200 are the default address and port.

Overall, I've read that dealing with netdata custom plugins can be tricky, but at least for the ones I am using it is as simple as it can get.

 

Great tutorial. I've just managed to get the base Netdata working on a test VPS

@EckyBrazzz I had to create a firewall rule before I could access the netdata dashboard as per the instuctions @ CSF Firewall - CentminMod.com LEMP Nginx web stack for CentOS .It might be worth mentioning this in your original just so you don't confuse noobs like myself