This application is comprised of three elements, an alert
collection process that runs on one server, an alert
generation (and deletion) tool that may be run on multiple
servers, and a console application (VT100 based) to
view the alerts from any server.
There is also a sample java GUI monitoring tool (because I like colours) that may
run as an application or a web page applet if you have java 1.4.x or above.
Together they provide a way of forwarding alerts from multiple servers to one point, allowing you to view alerts from any network connected Linux server that wishes to raise them in one place.
It is not a 'persistent' alert monitor. By this I mean that
if you stop and restart the server, all 'outstanding' alerts
it has collected up to that time will be lost. These should
not be very many unless your site has gone belly-up anyway.
The loss of alerts is deliberate. I have, and do still, work
with one of those alert forwarding tools that remember/store
alerts that cannot be forwarded at the immediate time;
if you have ever been paged at 2:00AM for a message three days old
you will understand the wisdom of chucking them away.
The central alert collector does keep an audit trail of all alerts recieved, deleted and acknowledged. Also each server records events it forwards, or tried to forward, so there is no loss of audittrail by throwing them away if the central collector cannot recieve them at that time.
It was written for my needs, so may not meet yours. Basically I have three linux servers,
only one I log onto regularly, the rest just do there important work. As such I missed an
error condition that caused me problems later on (a disk starting to die). This toolkit
will prevent that ever happening again as I now run the alert collector task on the server
I do log onto and have in my .bashrc a command to display the alerts each time I log on
via a tty session. My others forward the alerts to it so I see everything I need to.
Basically it works as
In the real world you probably have support staff, so they can use the vt100 interface to monitor the alerts in (almost) real time from any linux session; or if they have windows based PCs they can use the java GUI under windows. Alternately if you have a web server running on the same server as the alert collector they can use the java applet through a browser. I think I have all bases covered there.
It is important to note this is a toolkit, not an application, it provides a framework for raising, cancelling and automating alerts. You still need to do the hard work of writing scripts specific to your environment based on the samples provided.