Which simple stack to use for server monitoring?

Question

Monitoring

I also read about nagios, zabbix, cacti, tried to set up an ELK stack, all this works very well and works very reliably with the appropriate settings, but you need the simplest free (or cheap) service with auto-discovery of services installed on the server (SSH, CRON, Apache , Nginx, MySQL, Postgresql, php-fpm, redis, rabbitmq, etc. ) with log monitoring and custom alert - install agent and forget.
A vivid example of this for me is PM2 (Process Manager 2), but the price repels.

Is there a similar free solution?

Answer 1

You can look at https://my-netdata.io/
But in general, monitoring out of the box is IMHO unrealistic. The maximum will be seen that the hosts are alive and the services are running, but this does not mean anything.
The service can work, but fall out of the cluster. Or maybe on the contrary, it is so conceived that right now this instance is not synchronized.
The code may return 500 for invalid requests and this can be ignored. Or vice versa, if 500 are issued on some paths, then you lose money and you need to urgently notify everyone.
A response delay of up to 10 seconds may not be critical for you. And it can be critical if the 95th percentile went out for half a second.
I am generally silent about the database, a bunch of parameters that may or may not be important to you.