Director Background Service in two Server Cluster - Can it cause problems when running twice?

Hi there,

I’m building up a Icinga2 / IcingaWeb2 Cluster at the moment.

For IcingaWeb2 I got a two server cluster (behind loadbalancer) with shared volumes (/etc/icingaweb2 & /usr/share/icingaweb2/) to ensure same versions, sources, modules and its configuration.

The webinterface is NOT running on the same hosts Icinga2 is - so config-master of my Icinga2 setup is somewhere else (just for info in case anyone is asking).

Both nodes share the same database (icingaweb2 & director) which works fine so far.

At the moment there is running one director background service on each host.
They are recogniced by GUI for both hosts (both nodes show both background services).

So far, so good - but I wondered if this could cause any troubles when I decide to define import sources / syncs / jobs for director to import stuff automatically.

I didn’t find “anything” in the docs regarding this so I asked myself:
(1) Is there a internal mechanism that will ensure this kind of jobs will never executed in the same time from both daemons
=> Since there is a database table for running daemons, theoretically they could know there is a second one
=> If not - is there anything else that will ensure things won’t get messy when running two daemons (with same database)

(2) Is there a way to configure the daemon to place it’s pid to a specified location
=> “icingacli [director [daemon]] --help” indicates there’s no such option (since there is no info about it)
=> If it would be possible it would make things easy - dropping the pid file into the shared voldume, so only one node is running the daemon
=> Otherwise I could do it “manually” within the service file and ensure it is running only once - but I would prefer the daemon could do it by itself

(3) Clould it cause any other side effects running the daemon ony once (one node without it)
=> Doc indicates daemon is needed for exactly these things I try to run only once (+ Housekeeping)
=> I guess Housekeeping is ment as DB-Housekeeping !? So I don’t care if this is something only one instance is doing
=> So I would think this shouldn’t be a problemm to have one node without running a daemon locally
=> Is there anything I’m missing in my thougths about running it only on one cluster node?

(4) If this all is just a bad idea - I could:
=> Install IcingaWeb2 with director module onto a third node (or even directly on my Icinga2 config-master)
=> Proxy /icingaweb2/director to the single running director instance
=> Only one daemon is required sice director gui only exists once
=> Not tested yet at all - so no idea this is running well (or at all)
=> Not a fan of my own thought … :smiley:

Above I wrote >I didn’t find “anything” in the docs regarding this< - this is (as you probably know) not quite true.

Of course I fond some lines like “Jobs are currently not executed in parallel” - but I guess this is ment like “one daemon will not execute two different jobs at the same time”.
Maybe it’s simple as that and it will also ensure two daemons will not execute the same job at the same time - but I don’t think this is part of the meaning.

So is there anyone with experience in setup director in a cluster setup?
Anyone with deeper knowledge how these things will be handled by the daemon?

I would appreciate any constructive thoughts about my thinking / plans for this setup.

Thanks in advance,