Nagios Support Forum

Don't ask me how this just happened to come up exactly one year TO THE DAY later, but the bosses are asking about the Metrics component again today. I had to dig up the old thread (https://support.nagios.com/forum/viewto ... it=metrics), and happened to notice that this all was literally a year ago today, lol. So, any progress on this? I checked out that feature request and saw that it's been closed as of September 28th this year with this as the last comment:

Next on the list is adding custom commands/metrics that the user will be able to customize what will display in the tabs and then this component will be as useful as the user wants it to be.
I have added a Feature Request to add customized Metrics in the Nagios TPS as task ID 6507.

That is one hell of a tease, is this glorious rework to the metrics component something that we might see in the near future? Pls?

Also, in XI5 is there a reason I can see a time period option under Metrics, but my boss can't? He's configured as a User currently, but I just tried giving him admin rights to see if it showed up then and it didn't. Still works for me, though. This is what I'm referring to....ok I just went back for screenshots to compare the 2 and now I'm not seeing a time period option either, but I swear I'm not crazy. It was there... I swear right to the left of "Limit to: [HOSTGROUP]" there was an option to specify when (i.e. last today, last 24h, last week, etc.). Please tell me I'm not crazy. If that wasn't there and I am just nuts, is there anyway something like that could be added. It'd be great to be able to check the metrics for say, "that CPU spike from yesterday at noon".

Metrics just show the last grabbed perf data, so time period wouldn't matter.

Ah, rats. So I am crazy then, meh. And yeah, I know that's how it currently works. What I'm wondering is if it'd be possible to check the metrics from a selected point of time. Since it's basically a snapshot of a single time I don't see a way this would work for a time period, but if there were an option to check a specific day/time that would be awesome. The reason we'd like this is because, say there's a CPU spike on one of our global servers at 10 pm tonight. Tomorrow when I come in I'd like to check the CPU usage for that time on all the virtual hosts that live on that global to see which one of them caused the spike. Currently the only way to do that is by analyzing graph data for each host individual which can take considerable time depending on how many virtual hosts there are on any given global.

Can't you just create one multistacked graph with all the CPU's you want to monitor for the past 24 hours and add that to dashboard? Only issue there though is when you need to add a new CPU to the graph, you have to recreate the whole thing. There needs to be an easier way to do that.

Nah, tried that already. You're limited to 5 checks you can add to the multistacked graph (unless I'm doing it wrong or something), and most of our globals have more than 5 virtual servers on them.

Edit: The 5 limit thing seems to be a limitation of 2014 (our prod server isn't XI 5 yet), XI 5 seems to allow more than 5. This could work, however our virtual hosts move around frequently enough that I'd be forced to create a lot of new multistacked graphs often because of the reason you already stated.

Quite a bit going on in this thread!

For Feature Request 6507 I see no movement, but it's less than a month old. 4243 had some work done, quite a lot actually, but that was back in May. I might need the developers to hop in and give an update directly, they might be working on something independently of there being a Feature Request filed for it.

Any updates from the devs?

FR 6507 is to create a customizable way to include metrics from a number of different check commands, command output and perfdata. This was added in anticipation of the original metrics changes not being able to guess all the different componenets/agents/commands that would have to be accounted for. Currently, it will only display metrics for known agents using the default commands and perfdata format so the parsing will work correctly.

I think you were originally talking about choosing a timeframe for the highcharts graphs in the 'Graphs' tab where you can choose 'last 4 hours' to 365 days ago. This is useful for visualization only since the graphs are being drawn and the timeframe being adjusted. Sadly this isn't necessarily doable with the Summary and Gauge tabs unless we retain historical data for each metric you want to track. Then again this information might be saved somewhere else for us to be able to access, but would take some testing.

This would be an entirely new feature request and you would have to see if it would get traction. Something along the lines of "Allow Metrics Component to show metrics from a select-able timeperiod" and then once i figure out if I can pull perfdata from the XI server or if I would have to create a way to save the metrics data somewhere to be referred to on a wider timeperiod display.

At the very least create a task in tracker and we can have a record of it.

Sorry it took so long, been busy/sick. Feature request filed: http://tracker.nagios.com/view.php?id=772.