The dedicated server I'm looking after started to crash occasionally, I suspect because of overload, so I need some performance/resource based monitoring software, preferably with web interface, something like OpenNMS which I have tried, but did not like. The OS is Linux CentOS 5.3
P.S. There over 50 websites running on the server, if the monitoring software could show which one is consuming most resources that would be most helpful.
There are a lot of answers. I personally use Zenoss, but there's a big list here: http://en.wikipedia.org/wiki/Comparison_of_network_monitoring_systems
I would recommend to use 2 software to do the monitoring:
Those will do complementary task of live monitoring and historical resource usage. If you want you can integrate them to have one single source of alerts.
You should use nagios with nrpe plugins to check the state and availability of the web server.
htop for local monitoring and apache-top if you are using apache.
monit is a well-designed and modular resource monitoring and management software which you can monitor server processes and execute meaningful causal actions (eg. restart a process, send alert etc.) in error situations. It has also a friendly user interface which you can access via HTTP protocol.
It comes with lots of modules for different protocols and you can easily plug it to your server.
There is also a commercially supported version which has some more capabilities.
Nagios, as others have mentioned, is an open-source infrastructure for monitoring remote systems from a central server. Remote systems report "service" status via a simple message, and the central monitoring server either "actively" asks for that status via a direct connection or the remote system reports it "passively" e.g. over http. (The quoted words are Nagios terminology.) The central server is configured using a set of text files which define the remote hosts, services, schedule, etc.
This is a very brief description of what Nagios offers, however gives you an idea of the type of infrastructure to expect. Nagios' strength comes from the simplicity of the status message. A "service" is anything that reports a status in the form that Nagios can ingest. It can be written in any language, on any platform. Many extensions have been developed and are available from the Nagios library.
The commercial version is called Nagios XI, and it puts a nice front end on top of the configuration files, using a database to store the metadata and RRDTool-based graphing module to generate reports.
We've been using Nagios XI for about six months. There is definitely a learning curve, and the administrative GUI needs some work to handle large-scale installations, however we're happy with the cost-benefit.
There are a number of open source and commercial monitoring (and alerting) tools. Nagios is fairly popular but a pain to configure... this goes double for OpenTSDB. If you're running a business (sounds like it with the sites), I'd recommend investing in commercial software like Metrink or New Relic. They can be expensive, but worth it to help keep your business going.
You could try SeaLion. It is a cloud based Linux server monitoring tool. It functions really well on CentOS 5.1(I'm using it). You might want to check the python dependency though(2.6+). Once this is done it's a cakewalk. Installation takes just seconds and UI is clean and simple. It gives a bird's eye view of all the servers so will be of great use to you.
