[Zenoss-dev] New User: 100% custom monitoring

[Eric Newton]

Hi Maddog,


You can make or reuse an existing nagios-style plugin to grab your
performance data.  It won't be hugely efficient, but I don't know how
much data you have to fetch.

If you want really scalable fetching/parsing of http data, you will
probably have to write your own collector.  That's not as hard as it
sounds (I do it all the time ;-).

The threshold system does allow for complex data thresholding, but right
Zenoss doesn't have simple way to specify the relationship between
values on different devices.  If the information is all on one device,
it's pretty straightforward, although you'll have to write your own
threshold.

You can always monitor the RRD files outside of Zenoss and post events
when you see unusual values.  We have been doing this here at Zenoss
verify that all collectors are actively updating the RRD files.

-Eric

Maddog wrote:
> Hello,
> 
> I'm a new user, not even finished downloading yet.
> I've watched the demo, read some guides and browsed the forums a bit - the zenoss system looks and sounds great.
> 
> I intend to use zenoss for application monitoring of a product that uses multiple-server environments - Database, front end (administration) and load-balanced groups of web servers.
> 
> Each software component has HTML monitoring pages that produce metric data on the software component.
> 
> I'm wondering if it is possible to poll these monitoring pages, grep data fields, aggregate them, produce averages, thresholds, events, graphs and dashboard units for human monitoring.
> 
> Is it possible to have more than one logical grouping?
> We have GSLB site which consist of two or more server groups that reside in different geographical locations, is it possible to logically group servers like this:
> 
>> Europe Site
>>> London
>>>> Server A
>>>> Server B
>>>> Server C
>>> Paris
>>>> Server A
>>>> Server B
>>>> Server C
> 
> Is it possible to recognize a drop in requests (or some other event based on monitoring rules) in:
> a) a single server
> b) a server group in London/Paris
> c) the entire Europe site
> 
> Our webservers are apache-based. For a load-balanced environment - is it possible to send alerts based on comparisons of data from the servers?
> 
> For example:
> In the London site, one server is getting more requests than others in the group (lets say 60% more), is it possible to create an alert based on this deviation?
> 
> At the moment, all aspects of the system are monitored by hand-crafted monitoring scripts, I'm looking for a more centralized management solution.
> 
> Thanks.
> 
> 
> 
> 
> -------------------- m2f --------------------
> 
> Read this topic online here:
> http://community.zenoss.com/forums/viewtopic.php?p=15539#15539
> 
> -------------------- m2f --------------------
> 
> 
> 
> _______________________________________________
> zenoss-dev mailing list
> zenoss-dev at zenoss.org
> http://lists.zenoss.org/mailman/listinfo/zenoss-dev
>