Master - satellite - clients setup checks stay in status pending

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


    I am setting up a test environment for icinga2.

    1 Master with 1 satellite will serve about a number of servers which can only initiate the communication themselves (top-down config sync). --> Not working currently

    That same master will however serve a batch of servers directly (don't even need a zone here to be honest). --> Working fine

    I have been struggling all day trying to setup this seemingly simple Icinga2 configuration.


    This is where I'm stuck:

    All of my nodes stay on status "Pending" in Icingaweb2.

    I suspect the problem resides in my zones.conf, I cant figure out what however... Any help would be appreciated.


    This is what I see in the debuglog on my master:

    [2017-06-19 17:28:04 +0200] debug/ApiListener: Not connecting to Endpoint 'vbcentos01.local' because that's us.
    [2017-06-19 17:28:04 +0200] debug/ApiListener: Not connecting to Zone 'master-zone' because it's not in the same zone, a parent or a child zone.
    [2017-06-19 17:28:04 +0200] debug/ApiListener: Not connecting to Zone 'satellite-zone' because it's not in the same zone, a parent or a child zone.
    [2017-06-19 17:28:04 +0200] debug/ApiListener: Not connecting to Zone 'vbcentos2.local' because it's not in the same zone, a parent or a child zone.
    [2017-06-19 17:28:04 +0200] debug/ApiListener: Not connecting to Zone 'vbcentos3.local' because it's not in the same zone, a parent or a child zone.


    zones.conf from master:

    zones.conf from satellite:

    zones.conf from client:


    This is how my zones.d looks like on my master:


    vbcentos3.local.conf ( = client)

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

  • I managed to get this working.
    I followed the docs to the letter and replaced all "NodeName" entries by the fqdn.

    I thought these were just local aliases and only the assigned value mattered.

    I also noticed a parameter was missing in the client host config (vars.client_endpoint = name). That seemed to help for the client.

    I do have another question:


    Is it possible to do a top-down config sync setup, but without the master initiating any connection for the syncing of the config files?

    The way I would like to set this up is the satellite always initiating the connection towards the master. Once the connection has been made between the satellite and master the master can sync the files with the satellite.

    I know this works for the monitoring plugin output itself (it says so in the docs).

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

  • I do not think that is possible as the master automatically connects to the satellites when a change is made.

    Linux is dead, long live Linux


    Remember to NEVER EVER use git repositories in a productive environment if you CAN NOT control them

  • I just tested this with a software firewall in between, blocking all traffic from the Master to the Satellite, but allowing tcp/5665 from satellite to the master.


    Apparently the configs do sync down from the master, via the satellite to the client (satellite has to initiate this connection or it would happen).


    However nothing is getting monitored on my client (checks stay in status "pending").

    Check results running on my satellite are getting through to the master.


    Edit:

    The agentless checks work for my client, but not for the satellite.

    The local checks work for my satellite but not for my client.

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