I use quite a few monitoring tools, one for each "kind" of monitoring...

-For server uptime, stats, etc I use nagios

-For security monitoring, I mix the open source OSSEC + Snort: http://www.ossec.net

-For availability+integrity remote monitoring of my sites/domains I use sucuri: http://sucuri.net

I realize the question could be more specific: Are there any particular annoyances or great features about your monitoring solution of choice?

We have been using munin and nagios in various setups. Munin is great for the ease of plugin creation but the alerting capabilities are lacking. It also has a few fundamental flaws such as gaps in the graph when a server can not be reached.

Nagios otoh is the kitchen sink solution with extensive alerting functions, but the UI is cluttered and configuration/maintenance tend to be pretty nightmarish.

Since i'm employed by Netmon (netmon.ca), I tend to drink my own kool-aid and use it most. That being said, since most of the products mentioned above are free, i'm assuming price is a big concern, so I would recommend looking for the simplest product to meet your needs. There really aren't any standards in this industry in terms of how to present monitoring data to users in a predictable way, and some solutions are abysmally complex, while others are simple but barely show you any important data.

I kind of like the nagios/zenoss relationship model, where nagios provides a great open-source platform for monitoring and analysis, and zenoss provides a clear emphasis on simplicity, presentation and automation on top of that platform. I think you should check those out, and i'm interested to see how that relationship and that business model develops in the future as well.

If you do have a budget for your monitoring solution though, there are some pretty cool solutions out there, but i'm obviously biased, so i'll refrain from going on a long diatribe about it.

Nagios and I'm hating each bit of it. While it works OK but a complex configuration could be a nightmare. In our case it's a legacy from the early days and grown up organically, over a hundred machines that could be a real pain. To clean it I ended up writing a custom configuration tool in Python using templates where configuration information is stored in a database.

I'm a fan of Munin. It's super easy to write custom scripts for it in any language. For example, here's a script written in ruby to graph Passenger memory usage: http://gist.github.com/37525

Munin is only one component of a complete monitoring setup however (good for observing long term trends in resource consumption). I also use Monit to issue alerts if a service is down, and Pingdom to monitor external service availability. Has worked well so far.

I use monit, but honestly on my 256mb VPS (non-mission critical) it threw a ton of false positives. Granted, I configured it wrong. The web UI is pretty neat, albeit basic, and if you set it up to allow restarting processes that can be handy in a pinch. Just be sure to check your configs, otherwise Apache might be restarted every time a rails app gets launched due to memory constraints. ;)

As a disclaimer, I am definitely a neophyte sysadmin so I'm certainly not using the tools correctly. I ended up turning monit on just for checking if my processes are running instead of checking for memory issues. Seems to be running fine, and I haven't heard a peep out of it since I changed the passenger module configs.

I'm keeping an eye on http://www.cloudkick.com/ (the core feature is management), but it isn't flying yet.

I use Fiveruns for rails app monitoring, which doesn't seem to be on your list. That gives me an idea of how fast my app is running, where the bottlenecks are etc.

Then I have monit setup to keep an eye on server load, hd's, nginx, postfix + dovecot etc, and if monit fails to restart services for whatever reason, I have pingdom as my last line of defence, at which point I look into it manually.

http://serverdensity.com/ is another interesting newcomer.

too expensive

It's free for 1 server and £10 GBP (~$15 USD) per server per month for multiple servers.

(This is a product from my company)

you know what would rock?

A hypervisor for server providers that does these tasks in background (without the root server owner seeing it) and alarms the provider if a hd breaks down, temperature is too high, etc! Which can then automatically react.

I think there is a need for your service, but I think it's too expensive. I wouldn't use it. A server starts at 50€ per month, why spend 1/4 of it just for monitoring?

Yeah, exactly. $360 per year for two servers is really expensive.

Depends what you compare it to. If you're looking at the "free" products like Nagios etc, then of course it's more expensive. But Nagios and the like require a lot of time and effort to set up, have ongoing maintance and require their own server (no point installing on the same server you're monitoring).

It's therefore about the value the service gives you i.e. support, ease of use, fast setup, easy alerting (e-mail & SMS) and so on.

I'm sure you read through my comment in your "Please Review" thread, so just for the reference of others - http://news.ycombinator.com/item?id=658749

In short - if your model works, great, but I can easily see it not working due to the service not being positioned properly.

The Dude for realtime SNMP monitoring, plus some custom software on projectors. Logs by email + custom software on other servers.

Could do with a more comprehensive, more revealing monitoring solution, but unwilling to look at Nagios and found Zabbix quite closed looking and Zenoss a bit less pleasant than I was hoping.

We use http://www.opsview.org/

I've found that http://wormly.com works very well - no false positives and if you enable the "agent" monitoring you get all kinds of nice graphs about MySQL performance too.

Monit, but looking closely at M/Monit - check it out, looks sweet.

We use HP SiteScope for our servers and Orion Solarwinds for our Network.

+ SiteScope is pretty easy to set up and can be extended with scripting.

- Support since it's gone to HP has been pretty bad.

We use solarwinds to monitor our network infrastructure.

I like www.woozweb.com quite a lot

french language however

I use Nagios to monitor about 60 machines, but I don't trust the booger...

Nagios for alerts, Cacti for reporting.

Zabbix for servers Cacti for networks

Osmius

splunk

Ummm.... Nagios?

other being log files