Dashboards customizations

mlabbepg · Post by **mlabbepg** » Tue Jun 01, 2021 1:43 pm

As our host and service definitions are getting pretty close to how we want them, We started playing with dashboards a bit more.

We've checked the videos here.

While the features and customizations seems really awesome at first, it seems to lack some basic customization features that turns them pretty disapointing.

Are we missing something?

A couple examples:

1-Whenever we want to edit a dashlet/graph already added to a dashboard, there doesn't seem to be a edit function and it looks like we need to recreate the dashlet/graph from scratch then add it again/resize/reposition to the existing dashboard(s).

2-We would like to create dashboards for different products/services we develop, which shows standard metrics (cpu/ram/disk/network usage, response time, and other relevant service checks) for hosts involved in these products/services (and then make the dashboards available to team members developing and supporting these different products/services).

We can create a Server Status dashlet for a host group without problem, but it doesn't seem to be possible to create performance graphs based on host groups (that and #1 will make dashboards management a real pain on the long term).

3-Multistacked performance graph: it would make more sense, for example, to name a graph "CPU Usage" (unit appear on side) and only show the host names in the legend (bottom of graph), rather than naming each line "CPU Usage (hostname) [percent]" in the legend:

3-cpu_usage_current.png

(I'm also not sure why the scaling shows > 100%)
vs

3-cpu_usage_goal.png

4-Gauge dashlets would be nice to use for cpu, ram, bandwidth usage, but it only seem to work properly with percentage values, which it doesn't seem to be able to read from NCPA's memory usage standard results:

Code: Select all

OK: Memory usage was 79.40 % (Available: 1.65 GiB, Total: 8.00 GiB, Free: 1.65 GiB, Used: 6.35 GiB)

It does work with CPU usage (with warn and crit) since the percent datasource is available (unlike memory usage), although we still would like to replace "percent" text in the middle of the gauge (since the text value already shows %) with "CPU" or $hostname$ instead. It also shows max value of 104.5% instead of 100%, which doesn't feel right.

5-Performance graphs (memory, cpu and disk usage) would be great out of the box, if we could remove redundant values (or at least unselect them by default). If we deselect "available" and "free" values on memory usage graph, and then add the graph to dashboard, the graph is added to the dashboard with all four values (enabled) anyway.

5-memory_usage.png

6-using Metrics dashlets we can show stuff like Disk Usage for host groups, but it doesn't seem possible to sort them other than top/bottom % utilization:

6-disk_usage_current.png

6-disk_usage_goal.png

7-While not as fancy as performance graphs, metrics could also be used for memory usage (and works with host groups), but it seems to look for % utilization values which it doesn't seem to be able to find from rrd (NCPA) results.

ssax · Post by **ssax** » Wed Jun 02, 2021 12:36 pm

I don't think you're missing anything, it looks like you have a good grasp of the current functionality, we are always open to submitting feature requests to development if you have any ideas like you posted so thanks for posting them!

1. Currently dashlets are not modifiable other than allowing hiding the title/pinning them/adding or deleting.

2. Unfortunately, it's not currently possible to grab all performance graphs by hostgroup. I have previously submitted a feature request for this but it hasn't been implemented as of yet.

3. What version of XI are you running? I labbed it up in XI 5.8.3 and they max out at 100 on mine, although it is possible to get over 100% for checks that return greater than 100%.

4. I labbed this up in 5.8.3 as well and the gauges are working properly for the NCPA memory usage check, are you just saying that the percentage gauge is missing for it? Since the graphs are based off of the metrics from the plugin and the percentage is only listed in the output, not as one of the performance data metrics, there wouldn't be a gauge for it (if that's what you're referring to here).

5. I've submitted this feature request:

Code: Select all

FR: XI - Have dashlets remember selected/deselected metrics and leave them that way once dashified

6. I've submitted this feature request:

Code: Select all

FR: XI - Metrics Dashlet - Allow custom sorting

7. The metrics functionality is limited to specific services created by the wizards since the check/output/perfdata needs to be parsed in order to calculate the utilization.

mlabbepg · Post by **mlabbepg** » Wed Jun 02, 2021 2:47 pm

Thanks for your reply.

1. Please make this a feature request if not existing already. Honestly it's a huge pain to have to start from scratch (especially for graphs) whenever you want to change a small thing like timeframe, or add/remove a value on a multistacked performance graph.

3. We're running Nagios XI v5.8.3 under ubuntu 20.04.2 LTS.

The max value is dynamic depending on the displayed values (e.g. max drops to 60% if I zoom on a less cpu-intensive zone), but gets up to 150% when the highest peak on screen is 96.405%. On percentage graphs, I would expect max to stay at 100% all time (unless maybe if you happen to have >100% values for some reason).

4. The percentage gauge is missing from Service Status Details (performance graph tab), and trying to create something out of Metrics/Memory Usage returns "No matching data to display.", and multistacked graph only offers values for the existing 4 metrics (see code section below).

As for the other gauges, it would be nice to use the "used memory" gauge, but it's not so great with a fixed max value of 100 (max value should be extracted from existing "total" metric).

Here's our current check command (check_ncpa v1.2.4) and result from ncpa agent v2.3.1:

Code: Select all

/usr/local/nagios/libexec/check_ncpa.py -H 192.168.28.152 -p 5693 -t secret -M 'memory/virtual'

OK: Memory usage was 72.20 % (Available: 2.22 GiB, Total: 8.00 GiB, Free: 2.22 GiB, Used: 5.78 GiB) | 'available'=2.22GiB;;; 'total'=8.00GiB;;; 'free'=2.22GiB;;; 'used'=5.78GiB;;;

I checked the agent via the API GUI and I do see the percent value in there so I'm not sure why the agent is skipping this value in the results returned to check_ncpa. Obviously that would solve alot of our issues. Should I open a separate thread to work this out?

-edit- if I query directly the percent value, it works. It's just not included with all other results :

Code: Select all

/usr/local/nagios/libexec/check_ncpa.py -H 192.168.28.152 -p 5693 -t secret -M 'memory/virtual/percent'

OK: Percent was 80.30 % | 'percent'=80.30%;;;

Do regional settings happen to have any impact on this? (we're french canadian)

5. Thanks. I suggest you also add a "hide deselected metrics" toggle button (like the "hide title" eye icon) so it keeps the graph legends cleaner when more metrics are disabled than used.

ssax · Post by **ssax** » Thu Jun 03, 2021 2:45 pm

1. Created:

Code: Select all

FR: XI - Make dashlets modifiable, you currently have to start from scratch (especially for graphs) whenever you want to change a small thing like timeframe, or add/remove a value on a multistacked performance graph.

3. I'll lab it up with Ubuntu and see what I can find.

4. It's likely just the way the developers wrote it, I have created an NCPA issue for this:

https://github.com/NagiosEnterprises/ncpa/issues/783

5. Created:

Code: Select all

FR: XI - Dashlets - Add a "hide deselected metrics" toggle button (like the "hide title" eye icon) so it keeps the graph legends cleaner when more metrics are disabled than used

Please keep in mind that the decision to implement the enhancements is at the discretion of our development team.

Nagios Support Forum

Dashboards customizations

Dashboards customizations

Re: Dashboards customizations

Re: Dashboards customizations

Re: Dashboards customizations