Hi, I'm trying to setup a distributed setup which grew out of a remote client execution scheme.
So I have a master that for the moment issue checks to a bunch of some remote client. This machine has the default /etc/icinga2/conf.d, and a zone (and an endpoint with the machine IP address). For each node to monitor, there's a zone whose parent is the master zone defined on the master machine (this is populated by Puppet exported/collected resources). Some checks are applied by virtue of the Service "command_endpoint" (done through applies). This allows the load or disk checks to be executed on the node itself. The hostalive/ssh default conf.d/ configuration allows the master to ping and check ssh those remote nodes. This is working correctly, and is mostly based on Walter Heck presentation material available on the Icinga website (thanks Walter!). Now, I have some nodes that aren't directly accessible by the master machine, but are by one of the already monitored mode (let's call it a.domain.com). So I've created a new zone on the master called "satellite-a.domain.com" with an endpoint containing the dmz ip address of this host. This zone has parent master. Then I created a zone for each internal hosts, whose parent is the "satellite-a.domain.com" zone. Those zones are also on the master. So we should have something like: master -> satellite-a.domain.com -> internal node This internal node is visible in the icingaweb UI, but not in the "inciga2 node list". All the ping/ssh checks for this internal node are critical, because they are executed from the master not from the satellite-a.domain.com. The "remote" checks (like disk, load) are not executed since there's no direct connectivity between master and the internal node. Nothing is relayed through the satellite-a.domain.com. Even if I manually change /etc/icinga2/conf.d/services.conf ping definition to add a "command_endpoint = a.domain.com", the ping node source doesn't change. I'm sure I've missed something important, but I fail to see what. Can anyone point me in the right direction? Anyone would have a sample distributed setup (preferrably managed by the puppet module :) that I could get inspiration from? Thanks! -- Brice Figureau _______________________________________________ icinga-users mailing list icinga-users@lists.icinga.org https://lists.icinga.org/mailman/listinfo/icinga-users
