---

Subject:	I just found what I think is
Author:	[Not Specified]
Posted:	2016-12-13 07:02

I just found what I think is the problem's exact pattern : when creating or editing a graph report,I can add any report as long as I choose a device, component and graph.

but some graphs are not associated to components but are device-related, for example CPU utiliziation.

On my 4.2.5 SP671 system, I cannot add any of those non-component related graphs, only ones associated with components. I know it used to work on my old "3.x upgraded to 4.2.4", but my fresh 4.2.5 isnt taking those graphs.

This is compounded by the lack of logging/errors, so I do not know what is going on at all and why it's failing.

Any idea

---

Subject:	You are dead right! It
Author:	Jane Curry
Posted:	2016-12-13 14:13

You are dead right! It appears not to be able to find graphs for devices where it CAN find graphs for components. I have just opened a JIRA ticket -https://jira.zenoss.com/browse/ZEN-26438 .

A possible get-around is to use a MultiGraph Report. This is inherently more flexible but slightly more complicated to setup. Fundamentally, a Graph Report can gather together existing GRAPHS; a MultiGraph Report can combine existing DATAPOINTS, ie. provided you have gathered a datapoint value for a device into an RRD file, then you can create a new graph that puts together datapoints for devices and/or components in different ways.

You have to create one or more Collections, one or more Graph Definitions and one or more Graph Groups.

The Collections are similar to the Graph Report - you can supply a filter to select specific devices but you can also have a collection defined by membership of a DeviceClass, System, Location or Group. You can choose whether to include components of devices if you select a specific device. It's actually much more flexible if you are interested in device-level graphs.

Under Graph definitions, you create a new Graph - just give it a unique name - and then you add Graph Points - probably Datapoints but they could also be Thresholds in monitoring templates. The trick here is to look at ADVANCED -> Monitoring Templates and select the template that supplies the datapoint. It is the full name of the datapoint that you need so CPUUtilization_CPUUtilization would be valid for a Linux Server (where the first CPUUtilization is the name of the data source and the second CPUUtilization is the name of a datapoint in that datasource). The good news is that you can start typing the datapoint name into the dialogue box and it will automatically show you all datapoints that match what you have typed so far - then you pick it off the list.

The Graph Groups then allows you to combine Collections (where you are collecting from) with Graph Definition (the data you want to display). You can also choose for any group, whether all the data should go on one graph or have a graph per group element.

It's really quite neat and worth exploring.

Cheers,

Jane

Email: jane.curry@skills-1st.co.uk    Web: https://www.skills-1st.co.uk