Server monitoring - experiences

Question

I've been reading up on "mon" for server monitoring and have created a config that monitors Apache, MySQL and FTP which is great. At the moment, I'm just trialing on one machine, though in the future the plan is to deploy this to most servers that we either own or are supporting.

I'm curious to know if there's people out there with experience of running mon across several/many servers and whether they've hit any particular maintenance problems with mon. It seems to me like the only problem might be for example if I need to update all of the email addresses that alerts get sent to for example.

Thanks,

James.

Answer 1

We use Nagios to monitor a few hundred servers for typically around half a dozen services per machine (time, disc utilization, SSH/SMTP/HTTP greetings, some HTTP content, ping), including with escalation lists, a single place to modify the e-mail addresses, and service dependencies (if the server is not responding to ping, don't send notifications for every service that is down). We have a central Nagios server, and use an external monitoring service to alert us if our upstream is having issues.

I'm very happy with our monitoring setup.

For capacity planning, we use munin, which is very easy to set up. We run that on most of our servers installed locally.

Sean

Answer 2

monitoring systems have a sort of built in set of tasks that must be managed manually. Something like updating contacts is a good example of this. I recommend choosing a monitoring system based on a balance of modularity versus redundant data. I prefer nagios with vim as the admin interface and git for managing revisions.