Nagios Support Forum

abrist wrote:Just to verify if the issue is high charts, can you try unchecking the box "Use Highcharts for Perfdata Graphs" in Admin --> Manage System Config and then click Update Settings and check the graphs. Are they split into two graphs?

Turning off HighCharts did split it into multiple graphs.

Dang. I will open a bug report (task id = 4607) concerning percentage uom and counter uom existing on the same graph.

Where are you looking at the graphs? By default, it puts all the check's perfdata onto one graph... but if you go to Graph Explorer > Scaleable Performance Graph and select a host, you can drill down to the exact perfdata you want to see and create a dashlet based off that.

jomann wrote:Where are you looking at the graphs? By default, it puts all the check's perfdata onto one graph... but if you go to Graph Explorer > Scaleable Performance Graph and select a host, you can drill down to the exact perfdata you want to see and create a dashlet based off that.

I had not explored that yet and it is really cool and will certainly be useful in some instances however as I mentioned before, with hundreds of monitored DHCP ranges we would want to be able to simply select the service with the alert and view the graphs there without having to go into Graph Explorer for each.

Jtessaro wrote:
I had not explored that yet and it is really cool and will certainly be useful in some instances however as I mentioned before, with hundreds of monitored DHCP ranges we would want to be able to simply select the service with the alert and view the graphs there without having to go into Graph Explorer for each.

I agree. And the fact that percentages and counters have no business existing on the same graph unless you want them to. It definitely should not be default.

IMHO Highcharts needs to be re-written so you can create templates. This would fix a lot of these problems.

I have been using XI 2012 and am in the middle of trying out XI 2014. We also make use of the option of selecting a service and immediately seeing all the graphs for that service. I have also noticed this behaviour, however I am of the opinion that the default should be to not put any counters in the same graph at all unless the user specifies it, as there is no guarantee that counters have the same order of magnitude - i.e. a counter that changes from 10 to 100 is insignificant and pretty much invisible when in the same graph as another counter with a value in the thousands/millions, even though the relative change of the former counter was very large.

Agreed, to start this in the right direction, would an option on the service detail pages to separate out each track within an rrd, into individual graphs, help your situation? I think this is going to be something best done in small steps.

I think that would help in our situation, assuming that Nagios XI remembers the setting for services where it has been selected. A global default setting would also be useful, and may be easier to implement instead? I.e. a check box together with the "Use Highcharts for Perfdata Graphs".

Saved and global settings might have to be added at a later date, but something that would at least allow separation should be relatively straight forward. I'll put in a feature request for it.