What is on Your Mind?

Hi.. Looking for advice / feedback

Work in IT in an operational team, number of years ago, all monitoring was manual, vast checklists for unix checks, checking space, checking application processes, files etc. filling in spreadsheets etc.

I took some basic scripting courses in ksh and overtime built up my skills.
Eventually started scripting some of the manual checks and alerting to a global text file that was monitored instead.
A year later decided to go a step further as there was no money for commercial IT monitoring tools.

So I set up a freeware database on an old unix server, ran a webserver and Solaris.
Put all events in to tables and designed a relatively simple but effective web page as a front end, so alerts could be seen clearly in one view and alerts auto clear etc.

Everything is managed by scripts, monitors space, processes, server response, some web page response times, file flows, connects to a lot of remote servers to monitor critical metrics and alert.
Also have started putting in some capacity / performance monitoring using the db to record and Google charts for graphical representation ..

It would never replace a commercial tool for large companies, but I am wondering if there may smaller companies with say 50-100 servers that something like this may interest.
As its all pretty much ksh scripts on top of some of freeware I mentioned, it's not encoded / packaged so I'm not sure how to go about things even if there was interest.
There has been great feedback internally, especially as cost is extremely low, literally a basic server and my time.
We are now investing in commercial tools as company has expanded

Interested in your thoughts

There are free solution like Cacti or Nagios which can provide all you have mentioned for free.
Personaly i use Cacti, i have been able to demonstrate that it is working with all stuff required (alerting, capacity planing, equipment support).

It is also completely open source, so you can hack around it (it is php / mysql / apache).

A shell script based solution will be hard to implement to work on all unix/linux operating systems (it can be done, but it will require much more effort).
It will be a monster script at the end if you wish to support everything

That is why folks use SNMP and/or agent based monitoring with central system.

As it is i work mainly with AIX where a ksh (actually ksh88) is the systems default shell. If you want to test the application under this OS or need help making sure it runs there contact me. I'd volunteer to help you.

bakunin