Icingaweb2/Grafana/InfluxDB not pulling data correctly

This forum was archived to /woltlab and is now in read-only mode.
  • Hi


    I have configured everything correctly I think on an icingaweb2/influxdb/grafana instance, however data does not appear to be pulling correctly into the module as shown in the pics.


    As you can see, hostalive does have datapoints but the module is empty. I have tried through direct and iframe. I have yet to try proxy because I'm struggling to find where I adjust the CURL timeout, but I suspect the error would be the same regardless.


    Does anyone have any ideas please?


    Thanks

  • Hi


    I am using the latest builds/commits as pulled from GitHub yesterday.


    I'm fairly new to Grafana, so I am using a prebuilt dashboard called icinga2-influxdb-host from here: https://grafana.com/dashboards/1572. I'm not sure how to check the JSON for the data pull for the dashboard if that is the right method?


    I think this is what you wanted (attached)


    Influxdb.conf here:


  • edited my post above.

    If you use that, does it work for you ?

    The post was edited 1 time, last by sru ().

  • I do not see how that should matter.

    In Post 1 you show that the black screenshot has data.

    All that is left is to configure the graph with dashboard name panel id so that this image shows in the host details.

  • So I found the metrics portion.


    From what I can tell, it's meant to be dynamic like this from the prebuilt dashboard so that it can generate the data per host.


    I am assuming that the idea behind the prebuilt dashboard is that it should change as my host does and the service does.


    Maybe I need to find the dashboard builder and check with them?


    Or am I missing something completely? Do I need to be building custom graphs like you have per service I want to render?

  • Do I need to be building custom graphs like you have per service I want to render?

    yes, that is needed. As is written at the right side of the dashboard.

    As for the different hosts, for me the plugin handles to provide the hostname to the url.

  • So, i downloaded the latest module grafana for you and installed it.

    What was needed to get a hostalive graph:


    In grafana, import the dashboard base-metrics from /usr/share/icingaweb2/modules/grafana/influxdb.

    Below Configuration/Grafana Graphs add a new graph hostalive and use dashboard base-metrics and panel id 9:

    Make sure that the generic_host template has enable_perfdata=true.

    After that, i got my graph:


  • Thanks SRU.


    That worked. Either I'm missing something in the documentation or it's not clear but I don't recall coming across anything that says:


    "This module will show you graphs but you must configure them in the modules section to display." At least when I was playing around with PNP4Nagios it would automatically pull data, but I believe that PNP4Nagios has been superseded by things like Grafana/InfluxDB these days.


    I am curious how you worked out what panel number it was for the dashboard though? Not sure why it is 9 and not 3...


    This can be marked resolved.


    Now to get my other services displaying perfdata to be harvested!

  • I am curious how you worked out what panel number it was for the dashboard though

    That is very easy.

    In Grafana, click a panel title and select view. The panel will be displayed "fullscreen".

    Watch the url for the panel id then.


    This can be marked resolved

    Would you please do that at the top of the page by setting a mark in the checkbox below the subject.

  • It would appear that the best method, from what I can tell at the moment, is a new dashboard per host to then pull graphs from. I haven't played with this yet, but will look at it in the coming days.

  • Mikesch do we have a possibility to make that more clear, that you need to create your own dashboards and templates and nothing works "automagically" :-)

  • I mean, he provides example dashboards for influx and graphite which already *are* templated.

    Problem is that persons like to "just get it runnig in icingaweb2" without having a clue of grafana first.

  • I'm only asking to make it more prominent. I'm slightly annoyed with users who refuse to read, and as such I document such things to prevent further questions.


    I'll see if I can find the time to introduce a "troubleshooting" section in the docs.

  • From what I have been playing with, no you will need one for each host as the templates like base-metrics and icinga2-influxdb-host will adjust the panelID depending on the dynamically generated list of services.


    Take, for example, panelID 34 for my switch is a different panelID 34 for one of my servers. If I am reading the way this works correctly, then choosing something that is say, Load at panel 34 for my switch may actually be Procs for my server depending on the number of services on the server.


    That said, I'm still going to test this theory out unless someone can immediately tell me I'm wrong