NetWatcher is an Open Source network "health monitor". What? No, please, not another one! There are so many - Nocol, Netsaint, OpenNMS, Big Brother, Big Sister, Mon, ... So, why did I do that? The tools that I looked at had one or more of the following problems: - require special agent run on the monitored systems. You may not want to install not well audited software on your production machines. - resource consuming for the monitoring server. Forking a process to check every object is way expensive. - have configuration in a text file. While I wholeheartedly appreciate this unixish approach, it becomes impractical if you have to describe several hundreds of objects in one file. What NetWatcher does: - Uses HTTP as monitoring and management console. - Is written in Perl5 for maximum portability. - Avoids propriatry agents, tries to stay with SNMPd (but propriatry agents are possible). - Uses SQL database to keep both current status of objects *and* configuration. - Is designed to logically separate monitors that do actual probes on objects from the core that makes decisions and sends notifications. What NetWatcher does not: - Does not have event correlation (you cannot tell it to ignore web server failure if the router in front of it is broken). - Does not have display filters (you cannot ask it to show only specific groups of objects/statuses). - Does not have object grouping and users grouping (you have to assign responsibilities individually). - Does not include any historic data processing. At the time of this writing (2001-08-17) the project has entered "early beta" stage. Almost all planned features are implemented (see TODO file for the rest), and the thing runs in production environment on our site (performing 389 distinct checks on 203 objects, with frequency from two checks per minute to one check per 15 minutes) and it is reasonably stable. If you think the concept matches your needs, give it a try. http://netwatcher.sourceforge.net/ Eugene